Guided by the Interaction of Person-Affect-Cognition-Execution (I-PACE) model and a self-stigma framework, this study aimed to investigate relationships between cognitive and affective self-stigma and behavioral self-stigma, problematic use of internet (PUI), and problematic smartphone use (PSU) among people with substance use disorders (SUDs). It also examined mediating roles for affective self-stigma in the relationships between cognitive self-stigma and behavioral self-stigma/PUI/PSU. Using a cross-sectional design, 530 participants diagnosed with SUDs in Taiwan were recruited from a psychiatric center in Taiwan. Mediation models were investigated using the Hayes' Process Macro Model 4. Mediation analyses indicated that cognitive self-stigma was directly associated with behavioral self-stigma (p < 0.001), but not with either types of PUI or PSU (p-values ranging from 0.41 to 0.76). Affective self-stigma was directly related to behavioral self-stigma (p < 0.001), two types of PUI, and PSU (β = 0.24-0.30; all p < 0.001); cognitive self-stigma was indirectly associated with behavioral self-stigma (β = 0.53; 95  % bootstrapping CI = 0.46, 0.60), two types of PUI, and PSU (β = 0.20-0.25; 95  % bootstrapping CI = 0.08-0.14, 0.31-0.37) via a mediating effect of affective self-stigma. Findings support the I-PACE model in a self-stigma context. The findings also suggest that addressing affective self-stigma may help prevent or reduce behavioral self-stigma, PUI, and PSU among people with SUDs. Longitudinal studies are warranted to investigate over time relationships between self-stigma and PUI/PSU in people with SUDs.