Optimum Temperature for Elite Running Performance

by Will LePage

Contributors: Austin Cookson, Emmett Cookson, Derek Eidt, and Kevin Mayer
May 2011


At the highest level of competitive running, which includes elite road races, the Olympics, and professional track leagues, athletes and coaches seek to capitalize upon every variable affecting race performance. One of these variables is the ambient temperature during a race. This study hypothesizes that the optimum temperature for running performance will be near room temperature for short distances (less than 400m) and will incrementally decrease as race distances significantly lengthen. The method of this research is to match the fastest race times in history with the mean temperature for the date of each race. Since the collection of the quickest races of all time represents the epitome of elite running success, the associated temperature values indicate the conditions for which elite runners are most likely to attain maximum performance. In turn, coaches and athletes can use this information to pursue the best ambient temperatures for optimum running performance. The optimum temperatures for race day range between 72.6ºF and 49.4ºF for the male 100m and marathon, respectively, and 73.4ºF and 51.8ºF for the female 100m and marathon, respectively.

For the complete version of this paper, download the file at http://runjeffcity.org/archive/Optimum-Temperature-for-Running-Performance.pdf.


On graphs below: dots indicate average temperatures for all-time fastest races on record with available weather data; lines indicate standard deviation.


Women — Optimum Temperature for Running Performance
Optimum Temperature Women


Men — Optimum Temperature for Running Performance
Optimum Temperature Women


Two forms of data will be collected: race times and historical temperatures. Since the purpose of this study is to evaluate elite running performance, it will consider only the fastest race times on record. Elite running encompasses the highest level of athletic competition, including invitational road races and circuits, the Olympics, the World Track and Field Championships, and the highest level of track leagues. Representative of the elite level of running is the list of all-time fastest performances at Olympic distances. Several groups, most notable the International Association of Athletics Federations, publish comprehensive lists of the best running performances. Since the goal of this study is to match the most favorable temperatures with race distances to allow elite athletes and coaches to make better planning decisions, the selection of race distances derives from the events for the Olympics. These race distances are 100m, 200m, 400m, 800m, 1500m, 5k, 10k, and marathon. For a more accurate analysis, this study adds the half-marathon, which is longer than 10k but shorter than a standard marathon (42.2k). By adding the half-marathon, the analysis will not lack a gap between 10k and 42.2k. Although the half-marathon is not an Olympic-sanctioned event, it is still a very common and highly respected road race distance, especially in non-Olympiad years. Additionally, this analysis includes data collection for both the male and female top performances, since genders are split at all levels of competition.

The low-hanging fruit for this data collection is the lists of all-time best race performances: the International Association of Athletics Federations (IAAF) maintains a highly reputable database of running performances from every elite race across the globe. Two of the tables in the database are the top-times for male and female athletes at a selection of distances, including the Olympic distances and the half-marathon that this study utilizes (Castellini). To gather the race times, the top lists database from IAAF is simply exported to Microsoft Excel, a spreadsheet program, to await calculation.

The difficult data collection step is finding the temperature for each date that accompanies a race on IAAF's list of all-time best performances. For example, if the fastest 100m dash time took place on August 10, 2005, in Paris, then the weather data for Paris on the same day would match that race. Several online weather services maintain historical weather data, but the best service this study found is the Weather Underground ("Weather History and Data Archive"). This free online service matches the mean, high, and low temperatures for most cities and dates within the past 15 years. In the case that the weather database does not contain a specific city and date for a race, then that race entry is skipped and the next fastest race time is considered instead.


The advantages of using historic weather data and the all-time fastest race performances are numerous. First, the race times truly represent the epitome of running performance, the fastest running examples ever accomplished by mankind. Secondly, the temperature values given from historic weather data eliminate the subjectivity or anecdotal weather evidence (e.g. that was an unusually cold day, but the temperature is unknown) that may be associated with races.


The drawbacks to this data collection include the fact that only the highest levels of elite running performances are considered. All other levels of the sport, such as recreational and amateur running, are neglected. Consequently, the results of this research are most applicable to elite runners who comprise an extremely small percentage of the running demographic. Nonetheless, these elite athletes are also the ones most interested in gaining every advantage legally and morally possible, so the knowledge from this research will be most pertinent and applicable to the elite running community.

Another weakness in this data collection method is the neglect for the time of day. The race performances from IAAF do not include the time of day that the race occurred, but rather only the day. Furthermore, the historical weather record only includes the mean, maximum, and minimum temperatures for each day. Thus, the exact temperature at which each race occurred is unknown. Nonetheless, most races take place during the mid-morning and early evening, which typically coincide with the mean temperature. Thus, the temperature for each race is assumed to be that of the mean temperature for the race date.