Context: Increasing rice productivity is key to achieve rice self-sufficiency in sub-Saharan Africa (SSA) where current consumption surpasses local production, mainly due to low yield associated with sub-optimal manage ment practices. Good agricultural practices (GAPs) – considered as an integrated practices including soil, water, weed, pest, and disease management are critical in increasing farmers’ yields. However, there is a lack of comprehensive assessment of on-farm yield variation with GAPs across production systems and agroecological zones (AEZs) at the continental level. Objectives: The objectives of the study were to (i) quantify yield variation with GAPs in three production systems and (ii) identify major production factors causing yield variation. Methods: From 2013 – 2022, GAPs were tested on-farm in 987 fields across 34 sites in 20 SSA countries. Yield data from GAPs plots were compared with farmers’ yields obtained from an independent yield gap survey. Results: Yield with GAPs varied significantly (p < 0.001) across production systems and AEZs. Mean yields were 5.1, 3.9, and 2.5 t ha–1 in irrigated lowland (IL), rainfed lowland (RL), and rainfed upland (RU), respectively. Yield gain with GAPs averaged 0.7, 1.1 and 0.8 t ha–1 in IL, RL and RU; and was smaller in sites having higher farmers’ yields. Overall, 78, 87 and 88 % of the GAPs plots in IL, RL and RU, respectively, had higher yields compared with farmers’ yields. GAPs significantly (p = 0.01) reduced yield variation across production systems by 25, 29 and 20 % in IL, RL and RU, respectively. N, P and K use efficiencies, defined as partial factor pro ductivity (kg grain/kg nutrient applied), were significantly (p < 0.001) higher in IL (59, 153 and 151 kg grain/kg N, P and K, respectively), followed by RL (47, 123 and 129 kg grain/kg N, P and K) and lowest in RU (31, 81 and 80 kg grain/kg N, P and K), with positive correlations between yield and N, P and K use efficiencies. Across production systems and AEZs, bunding, levelling, basal N, P and K and total N rates were among the top ranked management practices influencing yield, where high yielding plots were associated with good levelling and bunding. Conclusion: There is substantial potential to further increase productivity by improving on-farm management practices—particularly to enhance nutrient use efficiency—to close rice yield gaps across diverse production systems in SSA. Significance: The study contributes to better understanding of the effect of GAPs on yield and yield variation, and production factors that influence yield variation at a large geographical area of SSA.