Detection of methicillin-resistant Staphylococcus aureus (MRSA) isolates exhibiting intermediate susceptibility to glycopeptides (GISA) is challenging for clinical microbiology laboratories. We compared three different screening assays for evaluating trends in decreased glycopeptide susceptibility during two periods. Ninety four and ninety five consecutive MRSA blood isolates from 189 bacteremic patients collected during periods A (1989-1994) and B (1999-2001), respectively, were screened in parallel for vancomycin or teicoplanin susceptibility by glycopeptide-containing brain-heart infusion agar (BHIA) tests, Etest MICs performed at a standard (0.5 McFarland) or high (2.0 McFarland) inoculum on Mueller-Hinton (MHA) or BHIA, respectively. Any MRSA isolate yielding <50 CFU (representing <10(-6) of the plated inoculum) on either BHIA containing 2 mg/l of vancomycin (V2-BHIA) or 5 mg/l of teicoplanin (T5-BHIA) was considered as fully susceptible to vancomycin or teicoplanin, respectively. The proportion of MRSA isolates yielding>50 CFU on either V2-BHIA or T5-BHIA significantly (P<0.01) increased from 7/94 (7.4 %) or 8/94 (8.5%) in period A to 16/95 (16.8 %) or 14/95 (14.7%) in period B, respectively. Vancomycin Etests MICs on MHA were of lower sensitivity (<30% and<65%), but high specificity (100% and 99%) on periods A and B isolates, respectively, compared to those on V2-BHIA. Vancomycin Etests MICs on BHIA were of a higher sensitivity (57% and 81%) but lower specificity (91% and 65%) compared to those on V2-BHIA, on periods A and B isolates, respectively, reflecting an unexpectedly high number of false positive isolates on period B isolates (28/95). Screening of decreased susceptibility to vancomycin or teicoplanin on V2-BHIA or T5-BHIA, respectively, may represent simple low-cost alternatives to Etest MICs, minimising the risk of missing potential GISA isolates.