A foundational strategy to promote implementation of evidence-based interventions (EBIs) is providing EBI training to therapists. This study tested an online training system in which therapists practiced observational coding of mock video vignettes demonstrating family therapy techniques for adolescent behavior problems. The study compared therapists ratings to gold-standard scores to measure therapist "reliability" (consistency across vignettes) and "accuracy" (approximation to gold scores); tested whether reliability and accuracy improved during training; and tested therapist-level predictors of overall accuracy and change in accuracy over time. Participants were 48 therapists working in nine community behavioral health clinics. The 32-exercise training course provided online instruction (about 15 min/week) in 13 core family therapy techniques representing three modules: Family Engagement, Relational Orientation, Interactional Change. Therapist reliability in rating technique presence (i.e., technique recognition) remained moderate across training; reliability in rating extensiveness of technique delivery (i.e., technique judgment) improved sharply over time, from poor to good. Whereas therapists on average overestimated extensiveness for almost every technique, their tendency to give low-accuracy scores decreased. Therapist accuracy improved significantly over time only for Interactional Change techniques. Baseline digital literacy and submission of self-report checklists on use of the techniques in their own sessions predicted coding accuracy. Training therapists to be more reliable and accurate coders of EBI techniques can potentially yield benefits in increased EBI self-report acumen and EBI use in daily practice. However, training effects may need to improve from those reported here to avail meaningful impact on EBI implementation.