Background: The popularity of endurance running events has rapidly increased in recent years with more recreational runners entering the field. How recreational runners train is not well known. Understanding this and the relationship between training and performance in this group of runners is important for prescribing appropriate training to maximise performance and decrease the risk of injury. This forms the underlying theme throughout this thesis. Aim: The broad aims of this thesis were to better understand the ad libitum training habits of well-trained competitive recreational runners and to determine the relationships between performance, training load, and submaximal heart rate (HR) in this cohort. Methods: Five inter-related studies were performed to: 1) determine relationships between 56-km race performance and pacing (n = 7,327) in competitive recreational runners; 2) determine relationships between 56-km race performance, pacing, and training load in competitive recreational runners (n = 69); 3) determine the accuracy of GPS sport watches in measuring distance (n = 255); 4) develop a feasible and reliable submaximal running test, and 5) determine relationships between performance on a submaximal running test, training load, and submaximal HR in well-trained competitive recreational runners (n = 29). Main findings: A group of well-trained competitive recreational runners performed 44 ± 22 km/week (median ± IQR) in a six-month time frame while training ad libitum. This group had a wide range of inter-individual differences in training load performed even when considering participants who had the same relative marathon performance. The same group of well-trained competitive recreational runners maintained most of their training over a sixmonth period in a range of 0.81 – 1.14 for the acute: chronic workload ratio (ACWR). When the ACWR values reached > 1.50, it was mainly due to participation in endurance running races (> 21-km). When looking at relative weekly changes in training load, the maximum increase was 30% with only two participants having maximum increases of < 10%. The increases in load were predominantly short term (one to two weeks). Submaximal HR had a negative linear relationship with performance in 21% of the study participants. In those participants, poor performances were associated with a higher submaximal HR. Training load was only related to changes in performance in one participant. Conclusion: This thesis confirms that no single variable can provide the necessary information on how to adjust training load to maximise performance. Athletes, coaches, and sports scientists need to have a holistic view of stress exposure and how this affects the body. Although we can only speculate, when the participants had a poor performance it may have been due to factors such as lack of motivation, fatigue, mental stress, dehydration, and/or sleep deprivation. It is important for runners, coaches, and sports scientists to approach the training load – recovery balance as being unique for each athlete. Even in a homogenous group of well-trained competitive recreational runners, their ad libitum training load is widely varied and was not associated with performance or ability level. The balance should be adjusted over time based on the athlete’s symptoms.