Classical galactosemia is an autosomal recessive disorder resulting from deficient galactose-1-phosphateuridyl transferase (GALT) activity. Verbal dyspraxia is an unusual outcome in galactosemia. Here we validated a simplified breath test of total body galactose oxidation against genotype and evaluated five potential biochemical risk indicators for verbal dyspraxia in galactosemia: cumulative percentage dose (CUMPCD) of (13)CO(2) in breath, mean erythrocyte galactose-1-phosphate, highest erythrocyte galactose-1-phosphate, mean urinary galactitol, and erythrocyte GALT activity. Thirteen controls and 42 patients with galactosemia took a (13)C-galactose bolus, and the (CUMPCD) of (13)CO(2) in expired air was determined. Patients with <5% CUMPCD had mutant alleles that severely impaired human GALT enzyme catalysis. Patients with > or =5% CUMPCD had milder mutant human GALT alleles. Twenty-four patients consented to formal speech evaluation; 15 (63%) had verbal dyspraxia. Dyspraxic patients had significantly lower CUMPCD values (2.84 +/- 5.76% versus 11.51 +/- 7.67%; p < 0.008) and significantly higher mean erythrocyte galactose-1-phosphate (3.38 +/- 0.922 mg/dL versus 1.92 +/- 1.28 mg/dL; p = 0.019) and mean urinary galactitol concentrations (192.4 +/- 75.8 mmol/mol creatinine versus 122.0 +/- 56.4; p = 0.048) than patients with normal speech. CUMPCD values <5%, mean erythrocyte galactose-1-phosphate levels >2.7 mg/dL, and mean urinary galactitol levels >135 mmol/mol creatinine were associated with dyspraxic outcome with odds ratios of 21, 13, and 5, respectively. We conclude that total body oxidation of galactose to CO(2) in expired air reflects genotype and that this breath test is a sensitive predictor of verbal dyspraxia in patients with galactosemia.