We investigated the variability of strength trained athletes' self-selected rest periods between sets of heavy squat training. Sixteen strength-trained male athletes (M age = 23, SD = 3 years) completed two squat training sessions 48 hours apart. Each training session consisted of five sets of 5RM squats, interspersed with self-selected interset rest periods. A Gymaware linear optical encoder collected kinetic data for each squat and temporal data for each interset rest period. The participants' subjective ratings of the experience were taken before (Readiness to Lift [RTL]) and after (Rating of Perceived Effort [RPE]) each set. Mean total rest time and mean power output differed significantly between sessions. For both sessions, interset rest period increased, and power output decreased between Sets 3, 4, and 5 (95% CI range [−101, −17]) compared with Set 1. In both sessions, RPE increased significantly in Set 3 compared with Set 1 (95% CI range = [0.68, 2.19]), while RTL decreased significantly from Set 3 (95% CI range [−2.99, −0.58]) compared to Set 1. Interset rest period and power output demonstrated fair reliability between sessions (mean intraclass correlation coefficient = 0.55), while RPE and RTL demonstrated good and excellent reliability, respectively (mean intraclass correlation coefficient = 0.63 and 0.80). In conclusion, highly trained strength athletes demonstrated a significant difference in their between session power output and total rest time when using self-selected interset rest periods, despite stability in their subjective ratings of fatigue and effort. Interset rest periods can be self-selected reliably to complete strength training in heavy squat protocol; however, power output may decline during the set.