A rotation experiment was conducted on an alluvial soil at Central Potato Research Station, Jalandhar, India between 1994 and 1999 to develop an integrated nutrient management programme for a potato-sunflower-paddy rice rotation. It tested 10 fertilizer treatments that combined application of 50, 100 and 150% of the recommended rates of inorganic nitrogen (N), phosphorus (P) and potassium (K) with farmyard manure (FYM) and in some cases zinc. Potatoes were the most nutrient-responsive crop followed by paddy rice and sunflower. The responses of potatoes to both P and K were greater in frost and late blight years. In these experiments potatoes required NPK at 150% of the currently recommended rate for maximum tuber production (i.e. they required 270 kg N, 52 kg P and 150 kg K/ha). Application of FYM at 30 t/ha with 270 kg of inorganic N/ha (FYM + N) was less effective than the use of NPK at 150% of the currently recommended rate. Application of K to replace that removed in the harvested crop was more effective in potatoes than giving K at 100 or 150% of the currently recommended rate. Sunflowers grown after potatoes that had received NPK at 150% of the currently recommended rate or with the FYM + N treatment required only half of the currently recommended rate of NPK for maximum yield (i.e. they required 23 kg N, 7 kg P and 13 kg K/ha), whereas paddy rice, grown as a third crop, required the full amount of NPK at the currently recommended rate (i.e. 120 kg N, 26 kg P and 67 kg K/ha). Application of K to adjust for its removal did not benefit sunflowers or paddy rice. Zinc had no direct effect on potatoes or any residual effects on the succeeding sunflower and paddy rice crops. The effects of the FYM + N treatment were cumulative, especially on potatoes which were the most responsive crop, whereas those involving the application of NPK at 150% of the currently recommended rate were not. Also, the NPK fertility of the soil was enhanced and more organic carbon was present following the FYM + N treatment than with NPK fertilizers applied at the currently recommended rate. Application of P resulted in a build-up of P but none of the treatments, even the applications of K to adjust for its removal, maintained the K status of the soil at its initial level.