Social anxiety is commonly assessed with self-report measures. This study aimed to investigate whether maximum anxiety levels during in vivo and virtual reality behavioral assessment tasks (BATs), and implicit approach-avoidance tendencies during the approach-avoidance task (AAT) explain more variation as predictors of daily social anxiety than self-report measures. A total of 62 university students (Mage = 20.79; SD = 4.91) with high levels of social anxiety completed self-report measures on fear of negative evaluation (FNE-B) as well as fear and avoidance in social situations (Liebowitz social anxiety scale-self report), in vivo and virtual reality BATs, and the AAT (independent variables) in the laboratory. On seven consecutive days, social anxiety, experiential avoidance, and negative social events (dependent variables) were assessed. The results revealed that fear of negative evaluation predicted everyday social anxiety and experiential avoidance. Fear and avoidance in social situations only predicted experiential avoidance. Neither implicit approach-avoidance tendencies during the AAT nor maximum anxiety levels during the in vivo and virtual reality BATs predicted any outcome variable. Our results support the use of self-report questionnaires in the assessment of social anxiety.