To correlate serum glycolic acid levels with clinical severity and outcome in ethylene glycol poisoning and to determine if glycolic acid levels are predictive of renal failure and the need for hemodialysis. We measured serum ethylene glycol and glycolic acid levels by gas chromatography/mass spectrometry for 41 admissions (39 patients) for ethylene glycol ingestion and performed retrospective chart reviews. Eight patients died, all of whom developed acute renal failure. Of the survivors, 15 also developed acute renal failure, whereas 18 did not. Of those with normal renal function, 8 had glycolic acid levels below detection limits (< 0.13 mmol/L) despite ethylene glycol levels as high as 710 mg/dL; 7 of these patients coingested ethanol. Pertinent initial laboratory data for each group are as follows (mean; range): Deceased: pH 6.99 (6.82-7.22); bicarbonate, 4.8 mmol/L (2-9); anion gap, 28.6 mmol/L (24-40); glycolic acid, 23.5 mmol/L (13.8-38.0); ethylene glycol, 136.5 mg/dL (6-272). Survived/acute renal failure: pH 7.07 (6.75-7.32); bicarbonate, 5.6 mmol/L (1-12); anion gap, 28.7 mmol/L (18-41); glycolic acid, 20.2 mmol/L (10.0-30.0); ethylene glycol, 238.8 mg/dL (12-810). No acute renal failure with glycolic acid > 1.0 mmol/L: pH 7.29 (7.12-7.46); bicarbonate, 14.7 mmol/L (4-23); anion gap, 16.5 mmol/L (10-26); glycolic acid, 6.8 mmol/L (2.6-17.0); ethylene glycol, 269.1 mg/dL (6-675). No acute renal failure with glycolic acid < 1.0 mmol/L: pH 7.41 (7.38-7.47); bicarbonate, 23.4 mmol/L (17-25); anion gap, 11.8 mmol/L (8-18); glycolic acid, 0.1 mmol/L (0-0.66); ethylene glycol, 211 mg/dL (8-710). The mean time postingestion to admission generally correlated with severity as follows: deceased, > or = 10.4 h; survived/acute renal failure, > or = 9.9 h; no acute renal failure with glycolic acid > 1.0 mmol/L, > or = 6.2 h; no acute renal failure with glycolic acid < 1.0 mmol/L, > or = 3.7 h. Hematuria was more prevalent than oxaluria (86% and 41%, respectively), but neither was individually predictive of acute renal failure. Good correlations were found between glycolic acid levels and anion gap (r2 = 0.7724), pH (r2 = 0.7921), and bicarbonate (r2 = 0.6579); poor correlations (r2 < 0.0023) occurred between ethylene glycol levels and glycolic acid, pH, anion gap, and bicarbonate. Measured ethylene glycol values were highly correlated with ethylene glycol values calculated from the osmolal gap (r2 = 0.9339), but the latter overestimates the true value by about 7%, on average. An initial glycolic acid level > or = 10 mmol/L predicts acute renal failure with a sensitivity of 100%, a specificity of 94.4%, and an efficiency of 97.6%. Ethylene glycol levels are not predictive of acute renal failure or central nervous system manifestations of toxicity. If only ethylene glycol values are available (measured or calculated), an initial anion gap > 20 mmol/L is 95.6% sensitive and 94.4% specific for acute renal failure when ethylene glycol is present. Likewise, initial pH < 7.30 is 100% sensitive and 88.5% specific for acute renal failure. We propose glycolic acid > 8 mmol/L as a criterion for the initiation of hemodialysis in ethylene glycol ingestion. Patients with glycolic acid < 8 mmol/L probably do not need dialysis, regardless of the ethylene glycol concentration, when metabolism of ethylene glycol is therapeutically inhibited. In the absence of glycolic acid values, an anion gap > 20 mmol/L or pH < 7.30 predicts acute renal failure.