High protein intake has been linked to increased type 2 diabetes (T2D) risk. However, if this association differs by protein from specific food sources, and if a habitual high protein intake affects insulin resistance and prediabetes risk are largely unknown. We aimed to investigate associations between protein intake from different food sources with longitudinal insulin resistance, and risk of prediabetes and T2D. Our analyses included 6822 participants aged ≥45 years without diabetes at baseline in three sub-cohorts of the prospective population-based Rotterdam Study. We measured protein intake at baseline using food-frequency questionnaires. Data on longitudinal homeostatic model assessment of insulin resistance (HOMA-IR), and incidence of prediabetes and T2D were available from 1993 to 2014. During follow-up, we documented 931 prediabetes cases and 643 T2D cases. After adjusting for sociodemographic, lifestyle, and dietary factors, higher total protein intake was associated with higher longitudinal HOMA-IR and with higher risk of prediabetes and T2D (per 5% increment in energy from protein at the expense of carbohydrate, for HOMA-IR: β = 0.10, (95%CI 0.07, 0.12); for prediabetes: HR = 1.34 (1.24 1.44); for T2D: HR = 1.37 (1.26, 1.49)). These associations were mainly driven by total animal protein (for HOMA-IR: 0.10 (0.07, 0.12); for prediabetes: 1.35 (1.24, 1.45); for T2D: 1.37 (1.26; 1.49)). The harmful associations of total animal protein were contributed to by protein from meat, fish, and dairy (e.g. for HOMA-IR: protein from meat, 0.13 (0.10, 0.17); from fish, 0.08 (0.03, 0.13); from dairy, 0.04 (0.0003, 0.08)). After additional adjustment for longitudinal waist circumference, associations of total protein and total animal protein with longitudinal HOMA-IR and prediabetes risk were attenuated, but remained statistically significant. Total plant protein, as well as protein from legumes and nuts, from grains, from potatoes, or from fruits and vegetables, was not associated with any of the outcomes. Higher intake of animal protein, from meat, dairy and fish food sources, is associated with higher longitudinal insulin resistance and risk of prediabetes and T2D, which may be partly mediated by obesity over time. Furthermore, plant protein from different sources is not related to insulin resistance, and risk of prediabetes and T2D. Our findings highlight the importance of specific protein food sources and that habitual high animal protein intake may already in early stages be harmful in the development of T2D.