Purpose: To investigate the relationship between physical performance tests and on-ice external load from simulated games (scrimmages) in ice hockey. Methods: A total of 14 players completed a physical performance test battery consisting of 30-m sprint test—run and 30-m sprint test—skate (including 10-m split times and maximum speed), countermovement jump, standing long jump, bench press, pull-ups, and trap bar deadlift and participated in 4 scrimmages. External load variables from scrimmages included total distance; peak speed; slow (< 11.0 km/h), moderate (11.0–16.9 km/h), high (17.0–23.9 km/h), and sprint (> 24.0 km/h) speed skating distance; number of sprints; PlayerLoad™; number of high-intensity events (> 2.5 m/s); accelerations; decelerations; and changes of direction. Bayesian pairwise correlation analyses were performed to assess the relationship between physical performance tests and external load performance variables. Results: The results showed strong evidence (Bayes factor > 10) for associations between pull-ups and high-intensity events (τ = .61) and between maximum speed skate and peak speed (τ = .55). There was moderate evidence (Bayes factor >3 to <10) for 6 associations: both maximum speed skate (τ = .44) and countermovement jump (τ = .44) with sprint speed skating distance, countermovement jump with number of sprints (τ = .46), pull-ups with changes of direction (τ = .50), trap bar with peak speed (τ = .45), and body mass with total distance (τ = .49). Conclusion: This study found physical performance tests to be associated with some of the external load variables from scrimmages. Nevertheless, the majority of correlations did not display meaningful associations, possibly being influenced by the selection of physical performance tests.