Repetitive negative thinking (RNT; e.g., worry and rumination) is common across emotional disorders, as is the tendency to generate negative interpretations (interpretation bias). Ameliorating negative interpretations via cognitive bias modification of interpretations (CBM-I) reduces worry/rumination, and improves mood in people diagnosed with generalized anxiety disorder (GAD) or depression. We investigated whether these findings generalize to high worry or rumination populations, irrespective of diagnosis, and whether effects are increased by enhancing emotional engagement with training with active generation of positive resolutions of ambiguity and imagery. Community volunteers with excessive worry and/or rumination, who were above clinical cut-off on anxiety and/or depression measures, were allocated to an active control condition (n = 54), interpretation training condition with prior activation of RNT (CBM_RNT; n = 54), or training condition augmented with positive outcome generation and imagery (CBM_ENH; n = 53). Interpretation bias, RNT, and mood were assessed before and following 10 Internet-based sessions completed within a 1-month period. RNT and mood questionnaires were also completed at 1-month follow-up. After training, both forms of CBM-I (vs. control) facilitated more positive interpretations and reduced negative intrusions during a worry task. At 1-month follow-up, anxiety, depression, RNT, and worry in the past week were lower in the CBM-I than control conditions, but not rumination or trait worry. Compared with standard CBM-I, the augmented form facilitated more positive interpretations, reduced negative intrusions after training, and reduced trait rumination at 1-month follow-up, but it did not augment effects on trait worry, anxiety or depression. Interpretation bias maintains transdiagnostic RNT and Internet-based CBM-I can reduce longer-term RNT. (PsycINFO Database Record (c) 2020 APA, all rights reserved).