Several studies have shown that therapists are generally biased concerning their performed helping skills, as compared to judges' ratings. As clients' ratings of therapists' performance are better predictors of psychotherapy effectiveness than judges' ratings, this study examined the validity and effectiveness of a helping skills training program at reducing novice helpers' self-enhancement biases concerning their helping skills, in comparison to their clients' ratings. Helping skills were assessed by three objective measures (a knowledge multiple choice test, a video test and a role play), as well as by a self- and peer-reported questionnaire. In addition, some performed helping skills' correlates (relationship quality, session quality, and helpers' therapeutic attitudes) were assessed both by helpers and their simulated helpees. Seventy-two sophomores in psychology participated to this study, 37 being assigned to a 12-h helping skills training program, and 35 to a control group. Helpers were expected to assess the aforementioned performed helping skills and correlates as being better than their helpees' assessments at pretest, thus revealing a self-enhancement bias. At posttest, we expected that trained helpers would objectively exhibit better helping skills than untrained helpers while beginning to underestimate their performance, thus indexing a self-diminishment bias. In contrast, we hypothesized that untrained helpers would continue to overestimate their performance. Our hypotheses were only partly confirmed but results reflected a skilled-unaware pattern among trainees. Trained helpers went either from a pretest overestimation to a posttest equivalence (performed helping skills and performed therapeutic attitudes), or from a pretest equivalence to a posttest underestimation (performed session quality and performed therapeutic relationship), as compared to helpees' ratings. Results showed that trained helpers improved on all helping skills objective measures and that helpees' perceptions of their performance had increased at posttest. In conclusion, helping skills training leads helpers not only to improve their helping skills but also to have more doubts about their skills, two variables associated with psychotherapy outcome.