<bold>Purpose: </bold>Previous studies comparing the surgical accuracy between computed tomography (CT)- and magnetic resonance imaging (MRI)-based patient-specific instrumentation (PSI) methods have produced contradictory results. The aim of this study was to determine which is the more reliable imaging method (CT versus MRI) for patient-specific total knee arthroplasty (TKA). CT-based PSI is hypothesised to have an advantage regarding the number of outliers.<bold>Methods: </bold>A total of 22 randomised controlled trials (RCTs), including 1749 TKA cases, were eligible for the meta-analysis. RCTs, systematic reviews and meta-analyses on this topic published in databases before September 2018 were identified by a literature search. The primary outcome was the number of lower extremities with greater than 3° of difference in alignment angles between the postoperative outcomes and target outcomes. The parameters calculated from the meta-analysis included risk ratios (RRs) and 95% confidence intervals (CIs). Additionally, the publication bias and heterogeneity of the studies were assessed.<bold>Results: </bold>The risk of femoral rotational outliers in the PSI group (RR = 0.48; 95% CI 0.24-0.98) was significantly reduced. Furthermore, subgroup analysis showed that the accuracy in the CT-based PSI group was significantly higher than that in the MRI-based CSI group (RR = 0.31; 95% CI 0.10-0.92).<bold>Conclusion: </bold>This meta-analysis shows that when performing TKA with PSI, preoperative CT is beneficial for the production of the PSI, resulting in a significantly lower proportion of outliers in femoral rotational alignment. CT should be the preferred choice for imaging when performing TKA surgery with PSI to obtain better femoral rotational alignment.<bold>Level Of Evidence: </bold>Therapeutic study (systematic review and meta-analysis), Level I.