This study aimed to investigate the association between individual and combinations of macronutrients with premature death, CVD and dementia. Sex differences were investigated. Data were utilised from a prospective cohort of 120 963 individuals (57 % women) within the UK Biobank, who completed ≥ two 24-h diet recalls. The associations of macronutrients, as percentages of total energy intake, with outcomes were investigated. Combinations of macronutrients were defined using k-means cluster analysis, with clusters explored in association with outcomes. There was a higher risk of death with high carbohydrate intake (hazard ratios (HR), 95 % CI upper v. lowest third 1·13 (1·03, 1·23)), yet a lower risk with higher intakes of protein (upper v. lowest third 0·82 (0·76, 0·89)). There was a lower risk of CVD with moderate intakes (middle v. lowest third) of energy and protein (sub distribution HR (SHR), 0·87 (0·79, 0·97) and 0·87 (0·79, 0·96), respectively). There was a lower risk of dementia with moderate energy intake (SHR 0·71 (0·52, 0·96)). Sex differences were identified. The dietary cluster characterised by low carbohydrate, low fat and high protein was associated with a lower risk of death (HR 0·84 (0·76, 0·93)) compared with the reference cluster and a lower risk of CVD for men (SHR 0·83 (0·71, 0·97)). Given that associations were evident, both as single macronutrients and for combinations with other macronutrients for death, and for CVD in men, we suggest that the biggest benefit from diet-related policy and interventions will be when combinations of macronutrients are targeted.