Besides variation within the HLA gene complex determining a major part of genetic susceptibility to Type 1 diabetes, genome-wide association studies have identified over 60 non-HLA loci also contributing to disease risk. While individual single nucleotide polymorphisms (SNPs) have limited predictive power, genetic risk scores (GRS) can identify at-risk individuals. However, current models do not fully capture the heterogeneous progression of asymptomatic islet autoimmunity, especially in autoantibody-positive subjects. In this study, we investigated the additional stage-specific impact of 17 non-HLA loci on previously established prediction models in 448 persistently autoantibody-positive first-degree relatives. Cox regression and Kaplan Meier survival analysis were used to assess their influence on progression from single to multiple autoantibody-positivity, and from there to clinical onset. FUT2 and CTSH significantly accelerated progression of single to multiple autoAb-positivity, but only in presence of insulin autoantibodies and HLA-DQ2/DQ8, respectively. At the stage of multiple autoantibody-positivity, progression to clinical onset was impacted by various non-HLA SNPs either as independent predictors (GLIS3, CENPW, IL2, GSDM, MEG3A, and NRP-1) or through interaction with HLA class I alleles (CLEC16A, NRP-1, TCF7L2), maternal diabetes status (CTSH), or a high-risk autoantibody-profile (CD226). Our data indicate that, unlike for GRS, the weight of distinct non-HLA polymorphisms varies significantly among individuals at risk, depending on disease stage and other stage-specific risk factors. They refine our previous stage-specific prediction models including age, autoantibody-profile, HLA genotype, and other non-HLA SNPs, and emphasize the importance of stratifying accordingly to personalize time-to-event prediction in risk groups, or for preparing or interpreting prevention trials.