Introduction Although previous studies have related occupational exposure and epicondylitis, the evidence is moderate and mostly based on cross-sectional studies. Suspected physical exposures were tested over a 3-year period in a large longitudinal cohort study of workers in the USA. Method In a population-based study including a variety of industries, 1107 newly employed workers were examined; only workers without elbow symptoms at baseline were included. Baseline questionnaires collected information on personal characteristics and self-reported physical work exposures and psychosocial measures for the current or most recent job at 6 months. Epicondylitis (lateral and medial) was the main outcome, assessed at 36 months based on symptoms and physical examination ( palpation or provocation test). Logistic models included the most relevant associated variables. Results Of 699 workers tested after 36 months who did not have elbow symptoms at baseline, 48 suffered from medial or lateral epicondylitis (6.9%), with 34 cases of lateral epicondylitis (4.9%), 30 cases of medial epicondylitis (4.3%) and 16 workers who had both. After adjusting for age, lack of social support and obesity, consistent associations were observed between self-reported wrist bending/twisting and forearm twisting/rotating/screwing motion and future cases of medial or lateral epicondylitis (ORs 2.8 (1.2 to 6.2) and 3.6 (1.2 to 11.0) in men and women, respectively). Conclusions Self-reported physical exposures that implicate repetitive and extensive/prolonged wrist bend/ twisting and forearm movements were associated with incident cases of lateral and medial epicondylitis in a large longitudinal study, although other studies are needed to better specify the exposures involved.