The paper describes some properties and problems related to Perceptual Evaluation of Speech Quality (PESQ) delay compensation mechanism and relations between time-scale modification and quality estimate. The evaluation of PESQ algorithm with two types of signal stretching (uniform resampling-based stretch and pitch-preserving stretch) was performed and the listening tests with human listeners were carried out. PESQ performance was also compared against the 3SQM algorithm. Experimental results indicate that performance of PESQ is not sufficient for high precision quality estimation of time-scale distortions.