Summary Background In 2011, the Department of Health advised that a two-stage test approach should be used to improve accuracy of Clostridium difficile infection (CDI) diagnosis. No specific test protocol was established at that time. Aim To compare clinical features of inpatient CDI cases identified by toxin enzyme immunoassay (EIA) with those identified as polymerase chain reaction (PCR) positive but toxin EIA negative. Methods During a six-month period (2011–2012), 2181 liquid faeces samples submitted to North Bristol NHS Trust were tested by EIA for both toxin and glutamate dehydrogenase (GDH). A total of 215 toxin or GDH EIA-positive samples were tested by Cepheid Xpert PCR assay; 128 clinically evaluable inpatients were grouped by test result, and their duration of diarrhoea and 14-day mortality compared. Findings Inpatients with a positive PCR but negative toxin EIA had a significantly lower 14-day all-cause mortality [11%; 95% confidence interval (CI): 4–23%] than patients with a positive PCR and positive toxin EIA test (37%; 95% CI: 19–59%; P = 0.01), and a smaller proportion of patients had prolonged diarrhoea (>5 days or unresolved at death: 19%; CI: 9–32%, vs 67%; CI: 45–84%; P < 0.001). A positive toxin EIA test was a significant independent predictor of death [odds ratio (OR): 4.7, 95% CI: 1.4–15.4; P = 0.01] and prolonged diarrhoea (OR: 8.6; CI: 2.9–25.6; P < 0.001), but a positive PCR (given positive GDH EIA) was not. Conclusion The clinical significance of a positive PCR result without a positive toxin EIA is questionable; such a result is associated with a significantly lower mortality and shorter duration of symptoms than patients with a positive toxin EIA.