For our quantitative analysis of conference-travel-related emissions, we collected data for major MR conferences from 2016 to 2025 (see methods in the Appendix): EUROMAR editions 2016 (Aarhus), 2017 (Warsaw), 2018 (Nantes), 2019 (Berlin, joint with ISMAR), 2022 (Utrecht), 2023 (Glasgow) and 2024 (Bilbao); International Conference on Magnetic Resonance in Biological Systems (ICMRBS) editions 2022 (Boston) and 2024 (Seoul); and ENC-ISMAR 2025 (Asilomar, California). These conferences lasted between 4 and 6 d and hosted between approximately 470 (ICMRBS) and 1100 (joint EUROMAR-ISMAR) participants; most of the EUROMAR conferences hosted ca. 600–700 participants.

A notable first observation is the geographical distribution of attendees, as illustrated for the case of EUROMAR 2024 in Fig. . The majority of participants at EUROMAR conferences, often exceeding 80 %, come from Europe. The conference location slightly alters the distribution: we systematically detected additional “local” participants, comprising about 20–30 participants affiliated to the institute of the organizers, as well as more participants from the hosting country. This trend is particularly pronounced for the 2024 ICMRBS in Seoul, e.g., where 200 of the ca. 580 participants were from the Republic of Korea (Fig. S1); for EUROMAR 2016 (Aarhus), 69 of the 620 were affiliated to a Danish institution, while this number was below 10 for all other EUROMAR editions. Likewise, ICMRBS 2022 (Boston) showed a strong attendance of participants from the Boston area (75 of ca. 470). This stronger inclusion of the local scientific community is, of course, a desired effect of moving the conference to different places. (We note that for ENC 2025 (Asilomar), this “local” effect is much less pronounced.)

Figure  shows cumulative distribution functions for all conferences (one-way distances). It illustrates that for conferences outside Europe, about half of the participants travel several thousand kilometers (round trip). Examples of travel-distance distributions are shown in Fig. c for EUROMAR 2024 (Bilbao) and Fig. e for ENC/ISMAR 2025 (California). At EUROMAR conferences, the most frequent distance traveled is between 800 and 1200 km, accounting for approximately 40 % of the participants. Another 25 % travel distances shorter than 800 km, and the remaining approx. 35 % travel distances longer than 2000 km, including long intra-European and overseas travel.

To translate this distance information into carbon emissions, assumptions must be made about the choice of transportation, as well as the respective per-kilometer emissions for trains and planes. (Note that we did not consider car travel for the European conferences.) Similarly to a study by , we assumed that attendees traveling for 400 km or less (one-way) choose to travel by train, while for longer distances, air travel is chosen. We also repeated the calculation for a cutoff of 800 km (one-way). In doing so, we explicitly used the actual distance of the train trip as calculated by Carbontracer . The 400 km cutoff, used in the previous study , is also supported by travel data from our institute: we reviewed several thousand trips of scientists traveling to/from our institute (data not shown), and found that a 400 km one-way distance is a reasonable cutoff above which air travel is chosen. For conferences outside Europe, we did not consider train travel as an option. For participants within 400 km of Asilomar or Boston, we assumed transport by car. We note that a more realistic metric would be to use travel times, rather than distances, as a criterion for travel preference. However, we refrained from such a metric because we lack the tools to calculate travel times efficiently for hundreds or even thousands of trips.

For the conversion from kilometers to metric tons of emitted CO2eq., we conducted a literature survey, paying particular attention to using realistic values and including indirect emissions (see Appendix for details). In brief, for air travel, one needs to take into consideration (i) the fact that short distances consume more fuel per kilometer than long-distance flights, (ii) radiative forcing and (iii) emissions related to infrastructure (airports). Likewise, for train transportation, we sought to obtain a holistic picture that includes not only the emissions related to producing electricity for propelling the trains, but also indirect effects. In particular, the construction and maintenance of railway lines and buildings are significant factors in train travel. Details on how we converted distances to emitted carbon are provided in the Appendix. In brief, a value of 25 g of CO2eq. per passenger kilometer in addition to the direct emissions is a realistic estimate for European countries. The carbon emissions for the production of electricity vary significantly for different countries (from 8 g CO2eq. (kW h)⁻¹ in Sweden to 594 g CO2eq. (kW h)⁻¹ in Poland . We have considered these differences; see the Appendix.

Our calculations have several shortcomings, which we shall list here. First, we do not know the mode of transportation chosen by each attendee. Moreover, for flights, we assumed direct flights from the airport closest to the participant's affiliation to the conference location. In reality, many journeys include connecting flights, which can significantly increase the carbon footprint by up to 100 kg . Thus, the estimated emissions due to air travel are likely underestimated by approximately 20 %. We also ignored possible car travel, which would slightly reduce the emissions (compared to flights) or increase the emissions (compared to train travel). Our estimates of emissions related to train travel are on the “pessimistic” side, and many websites of railway companies report lower numbers, usually because the indirect emissions are omitted. We explicitly want to be conservative here and avoid greenwashing of trains (see also below). We also note that the choice between train and air travel will depend not only on distance but also on the train connections and the availability of night trains. Finally, we chose distance as the criterion in our analysis. We note that the time of travel can also be used, especially for conference locations where the variance in time of travel for a fixed distance is large (e.g., due to high-speed rails connecting some cities but not others). However, using time as a criterion has the disadvantage of excluding night trains, which for a given distance typically takes more time. Technically, it is more challenging to estimate the travel time accurately than to calculate the distance precisely, and we thus use distance as the criterion. We note that a survey conducted among conference participants would be helpful to generate more accurate data. Such a questionnaire could be part of the conference organizers' feedback collection process.

Figure  shows the total and per-participant travel-related emissions of the analyzed conferences. The total travel-related carbon emissions of EUROMAR conferences were of the order of 700–900 t total or about 1.2–1.3 t CO2 per participant (assuming that flights are taken for distances longer than 400 km) or approximately 1 t (assuming that trains are taken up to 800 km). The per-person carbon footprint of conferences outside Europe is about 2 to 3 times higher due to longer distances traveled and also the less widespread availability and use of train travel for long-distance travel in, e.g., the USA.

In addition to travel, accommodation, catering and the conference site lead to further carbon emissions. Data about emissions of hotels have been collected, e.g., by the Cornell Hotel Sustainability Benchmarking (CHSB) Index or via the greenhouse gas conversion factors published annually by the UK Department for Environment, Food & Rural Affairs (DEFRA) or the French ADEME, and are available via web servers such as the hotel footprinting tool (https://www.hotelfootprints.org/, last access: 15 May 2025). The emissions scale roughly with the standing of the hotel due to the larger space for higher-rated hotels. For a 3-star hotel, they are in the range of 10–20 kg per night and person in a European country, which amounts to several tens of kilograms for an entire EUROMAR stay.

Meals can be estimated to produce approx. 5.6 kg CO2eq. for a meat-based meal, 3.8 kg CO2eq. for a vegetarian meal and 2.9 kg CO2eq. for a plant-based diet , which sums to approx. 14–28 kg CO2eq. for a 5 d conference. One can certainly debate whether meals should be counted as conference-specific, as they replace the ones the participants would have consumed if they were not attending the meeting.

The conference venue requires electricity and possibly natural gas for heating. Although we do not have precise values for the venues of previous EUROMAR conferences, one can estimate the corresponding carbon footprint to be approx. 10 kg per participant for the entire conference .

Overall, we estimate that the GHG emissions associated with conference attendance, excluding travel, amount to several tens of kilograms CO2eq.. Thus, transportation to the conference site, with typically more than 1 t of CO2eq., is the main contributor to the overall footprint of conferences.

One possibility for reducing the carbon footprint is to choose the mode of transport wisely. To establish a solid quantitative basis for comparing trains and planes, we have conducted a review of the relevant literature. To explicitly avoid any greenwashing of trains, we have considered not only the energy required for the transport itself (direct emissions), but also emissions related to building and maintaining the infrastructure. Moreover, we have explicitly considered the energy mix of the electricity grid (country-dependent), the actually traveled distance (which is most often longer by train than by plane as the trajectory is bound to the railway grid), the infrastructure-related emissions for construction, maintenance and operation (railway network, buildings, etc.), and typical passenger occupation of trains (including night trains) and planes. Various sets of assumptions, along with model calculations based on these assumptions, are shown in the Appendix. Figure  shows estimated CO2eq. emissions of train and plane travel for a set of cities in Europe. As examples, we have chosen journeys from a few European cities (Frankfurt, Vienna, Paris, Lyon) to the upcoming locations (at the time of writing) of EUROMAR (Oulu, Gothenburg, Lyon). In the case of Oulu, we have also accounted for the ferry transport between Stockholm and Turku. For all these cases, train travel emits much less CO2eq.. The reductions range from about 90 %–95 % (i.e., a 10–20-fold reduction) to about 75 % (4-fold reduction). We note here that for air travel, we have assumed direct flights; stopovers add around 100 kg CO2 for an additional takeoff .

In our estimations, we varied parameters such as the radiative forcing index (RFI) as well as the sources of direct emissions estimations (explained in detail in the Appendix). Figure  shows a representative and intermediate result of our analyses. To demonstrate the range of estimates we got by varying the RFI and other parameters, we present the “worst-case” and “best-case” scenarios in Fig. S2. In the “worst-case” scenario, taking a train “only” saves 40 %–80 % of CO2, while in the “best-case” scenario, taking a train instead of a plane saves up to 95 % of emissions.

These data demonstrate that at the individual traveler's level, the choice of transport is a meaningful way to reduce the carbon footprint. However, for longer distances, this is often not a viable option. Moreover, the emission reduction also tends to decrease for longer distances traveled, in part because train trajectories are longer than those of direct flights. We also note that, partly due to political choices, such as the tax exemption of kerosene, train travel tends to be more expensive and may therefore not be possible for this reason.

The significant variability in the per-person emissions of previous conferences (Fig. ) reveals that the location of the conference influences carbon emissions. Naturally, a more central location not only shortens the cumulative distance traveled but also increases the share of train travel. Based on data from previous EUROMAR conferences, we set out to predict the average carbon emissions per person for different conference locations (Fig. A). Choosing a central conference site, such as Frankfurt, Lyon or Vienna, can reduce the per-attendee emissions by a factor in excess of 2, compared to more remote locations like Northern Scandinavia or Israel. Flight distances to central conference locations are shorter and train travel is feasible for a larger number of participants. For a conference the size of EUROMAR, this means a reduction of approximately 500–800 t of CO2. Of course, besides the CO2 savings, it should be noted that colleagues working in places further away from the “center of mass” of the distribution are disadvantaged by systematically choosing more central locations, a factor that shall be considered.

We performed a model calculation for ENC, using the participants from the 2025 edition (which took place in California), and calculated the emission if the conference had taken place on the East Coast of the USA. We predict a ca. 25 % saving (400–500 t CO2 overall), which is due to the shorter distance for Europeans and possibly the density of NMR groups on the East Coast.

A possible solution to reduce the carbon footprint of conferences is to decentralize them by organizing events at several local hubs, which are connected virtually. Such a conference would then have several parallel sessions – which is common to EUROMAR, ENC and ICMRBS anyhow – and attendees would choose to listen to a talk happening locally or being streamed. A possible modality, presented previously , may look as follows. Initially, the organizers determine the event dates, select a primary location and identify several optional secondary sites. Next, they publicly announce the conference and main location and begin accepting participant applications. Then, taking into account the number and geographic distribution of applicants, participants are allocated to a final selection of venues in a way that minimizes total carbon emissions and maintains suitable attendance levels at each site. Finally, sessions proceed independently at each location, except during key plenary events, which are shared across all venues via videoconferencing. A case study of an international conference predicted this approach would cut emissions by one-third with three conference sites (one each in Japan, Europe and the USA) . present a similar strategy of dividing a conference into three hubs and connecting individual venues virtually, which could, according to their estimates, save up to 80 % of GHG emissions for a US-held conference.

We have performed a model calculation for a split conference to evaluate the potential reduction in carbon footprint. Figure B shows several model calculations for three possible parallel locations. Savings of the order of 25 % appear realistic, assuming that participants choose the closest hub. This number is similar to the one estimated in the study by .

Online-only conferences have a much lower carbon footprint. Carbon emissions of online meetings include the participants' devices, the internet infrastructure and the data centers. For example, Zoom/Google Meet/Teams video conferencing is estimated to result in ≈0.150 to 0.250 kg CO2eq. h⁻¹ per participant, which for an 8 h meeting per day amounts to 1.6 kg CO2eq. per person. This number is 2 to 3 orders of magnitude smaller than what we calculated for in-person meetings. A study of a large astronomy conference, for example, concluded that the online-only version reduced the carbon footprint by a factor of 3000; reported a 200-fold reduction for European political science conferences. Clearly, online-only meetings would result in a dramatic reduction in carbon emissions of MR meetings but could potentially be just as productive for advancing science and collaboration. One could argue that if an on-site conference produces 200 times more CO2, it should have at least 200 times better scientific output than an online conference. How exactly to measure the scientific outcome of a conference and how to weigh it against the GHG emissions are to be determined in future research. Reasonable metrics could be used to navigate the decision-making of conference formats.

Online conferences also have advantages: they can lift barriers to attendance for more junior scientists, scientists geographically distant from the “hubs” of research institutions and conference venues, and scientists at institutions with limited funding as well as parents and caregivers in general . Indeed, during the pandemic, for example, the attendance at the largest European meeting for geoscience rose by over 60 % . More affordable online formats may thus not only reduce carbon footprint by a factor of hundreds to thousands, but also bring us closer to equal opportunities for all groups of scientists.

We argue that an online format can actually be particularly useful for poster presentations. In-person poster sessions often suffer from noisy and crowded locations; often poster presenters wait for “clients”, presenters are not around or the posters are displayed only during part of the conference. Besides, printing and transporting posters is often cumbersome. Online formats, such as those explored by the Global NMR Discussion Meetings (https://www.globalnmr.org/online-conference/, last access: 15 May 2025), may provide a better experience.

An obvious way to reduce carbon emissions is to have fewer conferences and/or limit the number of participants per conference or, more precisely, the total distance traveled by participants for all the conferences they attend. Without compromising quality too much, one may achieve this by having meetings back-to-back at the same location. For example, a EUROMAR conference may be preceded by a more specific conference on small molecule NMR, for instance. This concept already exists in the form of satellite meetings that often take place before/after EUROMAR meetings. Similarly, we believe that attending a conference for its full duration – rather than leaving early, possibly to travel to the next meeting – is a meaningful way to improve the benefit-to-footprint ratio. Likewise, holding meetings biannually rather than annually may decrease the emissions by up to 50 %, as point out, or alternating between in-person and online editions.

We believe that it is often seen as a mark of success for a meeting to have as wide a geographical distribution of attendees as possible. In light of the need to reduce the carbon footprint, this view shall be reconsidered to adopt a climate-conscious mindset that conferences are best attended mostly by local scientists (local meaning within a country or a continent, which is not that local really), mixed with a small number of international scientists to foster and maintain cross-continental exchange.

A scenario along these lines with the highest reduction in GHG emissions would suggest that attendees who would otherwise need to travel long distances by plane could join online, while those able to travel with a low CO2 footprint by train would attend in person. Since a small number of long-haul flights often account for a disproportionately large share of total emissions, as previous work has shown , this strategy could substantially reduce the overall carbon footprint of conferences. Importantly, the experience for most attendees would remain similar to that of a traditional in-person meeting, as the majority could still gather on-site.

On a personal level, choosing to attend fewer conferences is one of the most direct ways to reduce one's carbon footprint. Such a choice, of course, comes with a careful evaluation of the benefits of attending a conference. Legitimate science-based reasons for attending conferences include staying up-to-date with the latest developments, networking with researchers and vendors, initiating collaborations, and increasing professional visibility. These goals appear especially important for early-career researchers seeking academic positions. For example, in a survey of doctoral students and postdocs in Germany, respondents indicated that conference travel had contributed to collaborative projects and publications. Notably, 8 % of postdocs reported having received a job offer as a result . Notwithstanding the potential benefits, studies show that beyond a certain threshold, increased travel does not correlate with higher academic performance . This study examined air travel emissions alongside scholarly metrics, such as the h index, and found no significant relationship between the volume of air travel and research output. This finding implies that while some conference participation can be professionally beneficial, excessive travel does not translate into greater academic success.