Despite a vast literature on one-leg hops and cutting maneuvers assessing knee control pre/post-injury of the anterior cruciate ligament (ACL), comprehensive and reliable tests performed under unpredictable conditions are lacking. This study aimed to: (1) assess the feasibility of an innovative, knee-challenging, one-leg double-hop test consisting of a forward hop followed by a diagonal hop (45°) performed medially (UMDH) or laterally (ULDH) in an unanticipated manner; and (2) determine within- and between-session reliability for 3-dimensional hip and knee kinematics and kinetics of these tests. Twenty-two healthy women (22.3 ± 3.3 years) performed three successful UMDH and ULDH, twice 1–4 weeks apart. Hop success rate was 69–84%. Peak hip and knee angles demonstrated moderate to excellent within-session reliability (intraclass correlation coefficient [ICC] 95% confidence interval [CI]: 0.67–0.99, standard error of measurement [SEM] ≤ 3°) and poor to excellent between-session reliability (ICC CI: 0.22–0.94, SEM ≤ 3°) for UMDH and ULDH. The smallest real difference (SRD) was low (≤ 5°) for nearly all peak angles. Peak hip and knee moments demonstrated poor to excellent reliability (ICC CI: 0–0.97) and, in general, moments were more reliable within-session (SEM ≤ 0.14 N.m/kg.m, both directions) than between-session (SRD ≤ 0.43 N.m/kg.m). Our novel test was feasible and, in most but not all cases, provided reliable angle estimates (within-session > between-session, both directions) albeit less reliable moments (within-session > between-session, both directions). The relatively large hip and knee movements in the frontal and transverse planes during the unanticipated hops suggest substantial challenge of dynamic knee control. Thus, the test seems appropriate for evaluating knee function during ACL injury rehabilitation.