There is conflict by evidence as to whether therapy with vancomycin plus an aminoglycoside is more nephrotoxic than therapy with either agent alone. Here we report the results of a prospective, non-randomized, open-label study of the incidence of nephrotoxicity in elderly patients who received vancomycin alone (32 patients) or an aminoglycoside alone (67 patients) or in combination (37 patients). The mean (95% confidence limits) incidence of nephrotoxicity, defined as an increase of ≥44·2 μmol/L in the serum creatinine, was 19% (5–32%) in patients receiving vancomycin alone, 24% (10–38%) in patients receiving vancomycin plus an aminoglycoside, and 12% (4–20%) in patients receiving an aminoglycoside alone (P > 0·05 for all comparisons). The corresponding absolute increases of the serum creatinine were 20·3±23·0 μmol/L (−24·8 to +65·4 μmol/L), 37·1 ±53·0 μmol/L (−67·2 to +140·4 μmol/L), and 22·1±31·8 μmol/L (−40·7 to +94·6 μmol/L). The absolute increase was significantly greater (P < 0·05) in patients receiving vancomycin plus an aminoglycoside than in patients receiving an aminoglycoside alone. A meta-analysis of seven previously published studies combined with our data revealed that the incidence of nephrotoxicity associated with combination therapy is 13·3±3·1% (7·3–19·4%) greater than therapy with vancomycin alone (P < 0·01) and 4·3± 1·4% (1·6–7·0%) greater than therapy with an aminoglycoside alone (P < 0·05). The clinical relevance of this finding may be limited in that the mean duration of antimicrobial therapy in three of the studies was greater than 21 days.