Higher variety of recommended foods, identified arbitrarily based on dietary guidelines, has been associated with better health status. Nutrient profiling is designed to identify objectively, based on nutrient content, healthier foods whose consumption should be encouraged. The objective was to assess the prospective associations between total food variety (food variety score, FVS) and variety from selected recommended and non-recommended foods (RFV and NRFV, respectively) and risk of chronic disease and mortality. In 1991–3, 7251 participants of the Whitehall II study completed a 127-item FFQ. The FVS was defined as the number of foods consumed more than once a week. (N)RFV(Ofcom) and (N)RFV(SAIN,LIM) were similarly derived selecting healthier (or less healthier) foods as defined by the UK Ofcom and French SAIN,LIM nutrient profile models, respectively. Multi-adjusted Cox regressions were fitted with incident CHD, diabetes, CVD, cancer and all-cause mortality (318, 754, 137, 251 and 524 events, respectively – median follow-up time 17 years). RFV and NRFV scores were mutually adjusted. The FVS (fourth v. first quartile) was associated with a 39 and 26 % reduction of prospective CHD and all-cause mortality risk, respectively. The RFV(Ofcom) (third v. first quartile) was associated with a 27 and 35 % reduction of all-cause mortality and cancer mortality risk, respectively; similar associations were suggested, but not significant for the RFV(SAIN,LIM). No prospective associations were observed with NRFV scores. The results strengthen the rationale to promote total food variety and variety from healthy foods. Nutrient profiling can help in identifying those foods whose consumption should be encouraged.