Background: Studies finding higher mortality rates for patients admitted to hospital at weekends rely on routine administrative data to adjust for risk of death, but these data may not adequately capture severity of illness. We examined how rates of patient arrival at accident and emergency (A&E) departments by ambulance—a marker of illness severity—were associated with in-hospital mortality by day and time of attendance. Methods: Retrospective observational study of 3 027 946 admissions to 140 non-specialist hospital trusts in England between April 2013 and February 2014. Patient admissions were linked with A&E records containing mode of arrival and date and time of attendance. We classified arrival times by day of the week and daytime (07:00 to 18:59) versus night (19:00 to 06:59 the following day). We examined the association with in-hospital mortality within 30 days using multivariate logistic regression. Results: Over the week, 20.9% of daytime arrivals were in the highest risk quintile compared with 18.5% for night arrivals. Daytime arrivals on Sundays contained the highest proportion of patients in the highest risk quintile at 21.6%. Proportions of admitted patients brought in by ambulance were substantially higher at night and higher on Saturday (61.1%) and Sunday (60.1%) daytimes compared with other daytimes in the week (57.0%). Without adjusting for arrival by ambulance, risk-adjusted mortality for patients arriving at night was higher than for daytime attendances on Wednesday (0.16 percentage points). Compared with Wednesday daytime, risk-adjusted mortality was also higher on Thursday night (0.15 percentage points) and increased throughout the weekend from Saturday daytime (0.16 percentage points) to Sunday night (0.26 percentage points). After adjusting for arrival by ambulance, the raised mortality only reached statistical significance for patients arriving at A&E on Sunday daytime (0.17 percentage points). Conclusion: Using conventional risk-adjustment methods, there appears to be a higher risk of mortality following emergency admission to hospital at nights and at weekends. After accounting for mode of arrival at hospital, this pattern changes substantially, with no increased risk of mortality following admission at night or for any period of the weekend apart from Sunday daytime. This suggests that risk-adjustment based on inpatient administrative data does not adequately account for illness severity and that elevated mortality at weekends and at night reflects a higher proportion of more severely ill patients arriving by ambulance at these times.