Background: Medium-chain acyl-CoA dehydrogenase (MCAD) deficiency is the disorder thought most to justify neonatal screening by tandem-mass spectrometry because, without screening, there seems to be substantial morbidity and mortality. Our aim was to assess the overall effectiveness of neonatal screening for MCAD deficiency in Australia. Methods: We identified MCAD-deficient patients from a total population of 2 495 000 Australian neonates (810 000 screened) born between April 1, 1994, and March 31, 2004. Those from a cohort of 1 995 000 (460 000 screened) were followed up for at least 4 years, and we recorded number of deaths and severe episodes, medical and neuropsychological outcome, and hospital admissions within the screened and unscreened groups. Findings: In cohorts aged at least 4 years there were 35 MCAD-deficient patients in those not screened (2·28 per 100 000 total population) and 24 in the screened population (5·2 per 100 000). We estimated that patients with this disorder in the unscreened cohort remained undiagnosed. Before 4 years of age, three screened patients had an episode of severe decompensation (including one neonatal death) versus 23 unscreened patients (including five deaths). At the most conservative estimate, relative risk of an adverse event was 0·44 (95% CI 0·13-1·45). In the larger cohort the relative risk (screened vs unscreened) of an adverse event by age 2 years was 0·26 (95% CI 0·07-0·97), also a conservative estimate. 38 of 52 living patients had neuropsychological testing, with no suggestions of significant differences in general cognitive outcome between the groups. Interpretation: Screening is effective in patients with MCAD deficiency since early diagnosis reduces deaths and severe adverse events in children up to the age of 4 years.