ChatGPT has emerged as a potentially useful tool for clinicians and the public in obtaining heath advice and diagnosis. In this study, six iterations of 22 pediatric emergency vignettes were entered into ChatGPT (total of 132 scenarios) to assess diagnostic accuracy, emergency call advice, and validity of advice given. ChatGPT correctly recommended contacting medical professionals in all cases but only advised calling emergency medical services (EMS) or 911 in 12 of the 22 scenarios. The correct diagnosis was made in 94% of cases, consistent with other research into ChatGPT. Considerably more research is required before ChatGPT could be recommended for diagnostic advice.