PURPOSE: To evaluate the effects of vocal therapy on voice disorders using different spectrographic parameters, before and after therapy. METHODS: This experimental retrospective study analyzed voice recordings of 67 dysphonic patients that had attended vocal therapy. Pre- and post-treatment spectrograms were analyzed by four speech-language pathologists. The following parameters were analyzed: spectrogram regularity, harmonic colors, spectrogram stability, presence of noise components, presence of harmonic and sub-harmonics. Data were submitted to statistical analysis, which aimed at identifying different patterns between genders and diagnoses. RESULTS: There was no difference between pre and post-therapy conditions for spectrogram regularity, however, most patients (58%) showed improvement. Harmonic colors remained stable (p=0.000). Significant improvement was observed, after voice therapy, regarding spectrogram stability (p=0.006), and presence of noise (p=0.007), harmonics (p=0.000) and sub-harmonics components (p=0.001). No relation was found between patient's gender and spectrographic improvement. Regarding diagnoses, differences caused by therapy were only significant for spectrogram regularity. CONCLUSIONS: Not all evaluated parameters showed significant improvements with therapy, however, acoustic spectrography proved to be an efficient tool to evaluate patients' progresses during vocal rehabilitation, complementing auditory-perceptual evaluation and composing a multidimensional assessment protocol.