This study conducted a Bayesian network meta-analysis (NMA) to compare the imaging and functional outcomes of patient-specific instrument-assisted unicompartmental knee arthroplasty (P-UKA), robot-assisted unicompartmental knee arthroplasty (R-UKA), and conventional unicompartmental knee arthroplasty (C-UKA). A comprehensive search was performed on five electronic databases and major orthopedic journals as of September 24, 2023. We included randomized controlled studies featuring at least two interventions of P-UKA, R-UKA, or C-UKA. Primary outcomes encompassed the deviation angle of hip-knee-ankle angle, as well as the coronal and sagittal plane alignment of femoral and tibial components. Secondary outcomes included patient-reported outcome measures (PROM), surgery time, revision rate, and complication rate. Bayesian framework was employed for risk ratio (RR) or mean deviation (MD) analysis, and treatment hierarchy was established based on rank probabilities. This NMA included 871 knees from 12 selected studies. In sagittal plane, R-UKA exhibited a significantly reduced deviation angle of femoral component compared to P-UKA (MD: 4.16, 95% CI: 0.21, 8.07), and of tibial component in comparison to C-UKA (MD: -2.45, 95% CI: -4.20, -0.68). Notably, the surgery time was significantly longer in R-UKA than in C-UKA (MD: 15.98, 95% CI: 3.11, 28.88). However, no significant differences were observed in other outcomes. Compared with P-UKA or C-UKA, R-UKA significantly improves the femoral and tibial component alignment in the sagittal plane, although this does not translate into discernible differences in functional outcomes. Comprehensive considerations of economic and learning costs are imperative for the judicious selection of the appropriate procedure.