Maximal strength can be predicted from the load-velocity relationship (LVR), although it is important to understand methodological approaches which ensure the validity and reliability of these strength predictions. The aim of this systematic review was to determine factors which influence the validity of maximal strength predictions from the LVR, and secondarily to highlight the effects of these factors on the reliability of predictions. A search strategy was developed and implemented in PubMed, Scopus, Web of Science and CINAHL databases. Rayyan software was used to screen titles, abstracts, and full texts to determine their inclusion/eligibility. Eligible studies compared direct assessments of one-repetition maximum (1RM) with predictions performed using the LVR and reported prediction validity. Validity was extracted and represented graphically via effect size forest plots. Twenty-five eligible studies were included and comprised of a total of 842 participants, three different 1RM prediction methods, 16 different exercises, and 12 different velocity monitoring devices. Four primary factors appear relevant to the efficacy of predicting 1RM: the number of loads used, the exercise examined, the velocity metric used, and the velocity monitoring device. Additionally, the specific loads, provision of velocity feedback, use of lifting straps and regression model used may require further consideration.