Semi-recumbent eccentric (ECC) cycling is increasingly used in studies of exercise with healthy and clinical populations. However, workloads are generally prescribed using measures obtained during regular concentric cycling. Therefore, the purpose of the study was to quantify the reliability of measures derived from a protocol that elicited peak ECC torque produced by the lower limb in a semi-recumbent position. Experiments were carried out on a dynamometer in a seated, semi-recumbent position identical to that of a custom-built ECC cycle, a modified Monark recumbent cycle. Thirty healthy participants completed two testing sessions. Each session comprised three series of six repetitions of a peak ECC torque protocol (PETP) on an isokinetic dynamometer. Absolute and relative reliability of peak torque, power, angle of peak torque, and work (recorded for each repetition) was determined using coefficient of variation (CV) and intraclass correlation coefficient (ICC), respectively. Ratings of perceived exertion (RPE), muscle soreness, and perceived effort (PE) were recorded pre-PETP, immediately post-PETP, and 1-min post each PETP. The protocol showed absolute reliability values <15% for mean peak (CV = 10.6–12.1) torque, power (CV = 10.4–12.3), angle of peak torque (CV = 1.2–1.4), and work (CV = 9.7–12.1). Moderate to high between-test relative reliability is reported for mean and highest torque (ICC = 0.84–0.95; ICC = 0.88–0.98), power (ICC = 0.84–0.94; ICC = 0.89–0.98), and work (ICC = 0.84–0.93; ICC = 0.88–0.98), respectively. Within-session peak torque, peak power, and peak work showed high relative reliability for mean (ICC = 0.92–0.95) and highest (ICC = 0.92–0.97) values. Overall, the PETP test provides a reliable way of determining peak ECC torque specific to semi-recumbent ECC cycling that may be used to prescribe workloads for this form of exercise.