Objective The purpose of this systematic review and meta-analysis was to evaluate the effect of electromagnetic field treatment on the symptoms of knee osteoarthritis (OA). In addition, the influence of the type of control group and other covariates have been investigated to identify the sources of heterogeneity in the results of the available clinical trials. Methods Randomized controlled trials reporting pulsed electromagnetic field–based therapies for the treatment of knee OA have been included. Main outcomes were self-reported pain and activity scores collected by Visual Analogue Scale (VAS) and/or Western Ontario McMaster Universities Osteoarthritis Index (WOMAC) at short term after treatment. Results Thirteen studies comprising 914 unique patients were included in the analysis. Overall reduction in pain score was observed after treatment (standardized mean difference −0.4059, P = 0.0091), while improvement in the activity score was not significant (standardized mean difference −0.4452, P = 0.0859). Type of control (i.e., placebo or alternative therapies) and time of follow-up resulted as the two major elements influencing the outcomes. Indeed, the restriction of the analysis to placebo-controlled trials demonstrated higher standardized mean differences between treatment and control groups, with lower P value for pain, while statistical significance became evident also for the activity score. On the contrary, no differences were observed pooling only studies comparing pulsed electromagnetic or magnetic fields to alternative treatments. In addition, longer follow-up correlated with lower differences between treated and control patients. Conclusions Pulsed electromagnetic field therapy effectively relieves knee OA symptoms at short term, but it is not superior to other conservative therapies such as physiotherapy.