Despite potato yield potential, its intensification level remain low in Rwanda, translating into low yield occasioned mainly by the decline in soil fertility. Field experiments were conducted in Birunga, Mudende [L1]) and Buberuka, Rwerere [L2] highlands Agro-Ecological Zones (AEZs), during September- December 2016 and March-June 2017 crop growing seasons to determine the effects of varying rates of N, P and K on potato tuber yield and yield components. The experiments were laid out using randomized complete block design with factorial arrangement, with four replicates. Factors were N rates (NX) i.e N1-0 kg ha-1, N2 – 50 kg ha-1, N3 – 100 kg ha-1 ; P2O5 rates (PX) i.e P1-0 kg ha-1, P2 – 50 kg ha-1, P3 – 100 kg ha-1and K2O rates (Kx) i.e K1- 0 kg ha1and K2- 50 kg ha-1. Number of tubers per plant, fresh tuber weight, small tuber yield, medium tuber yield, large tuber yield and total tuber yield were measured. Analysis of variance, performed using SAS-version 9.2, revealed that interaction effects of N×P×K were very highly significant on all parameters. Generally, N3×P3×K2 performed better than other treatments and recorded highest tuber yields in all situations: (32.73 ± 0.43) t ha-1[L1] and (29.36 ± 0.41) t ha-1 [L2] and (31.05 ± 0.52) t ha-1for pooled ANOVA. Contrarily to what happened at L2, N3P3K2 andN2P3K2 were not significantly different at L1. N2P3 K2 is recommended to L1 whereas N3P3K2 is recommended to L2.