Background: The 1918 influenza pandemic caused disproportionately high mortality among certain age groups. The mechanisms underlying these differences are not fully understood. Objectives: To explore the dynamics of the 1918 pandemic and to identify potential age-specific transmission patterns. Methods: We examined 1915–1923 daily mortality data in New York City (NYC) and estimated the outbreak duration and initial effective reproductive number (Re) for each 1-year age cohort. Results: Four pandemic waves occurred from February 1918 to April 1920. The fractional mortality increase (i.e. ratio of excess mortality to baseline mortality) was highest among teenagers during the first wave. This peak shifted to 25- to 29-year-olds in subsequent waves. The distribution of age-specific mortality during the last three waves was strongly correlated (r = 0·94 and 0·86). With each wave, the pandemic appeared to spread with a comparable early growth rate but then attenuate with varying rates. For the entire population, Re estimates made assuming 2-day serial interval were 1·74 (1·27), 1·74 (1·43), 1·66 (1·25), and 1·86 (1·37), respectively, during the first week (first 3 weeks) of each wave. Using age-specific mortality, the average Re estimates over the first week of each wave were 1·62 (95% CI: 1·55–1·68), 1·68 (1·65–1·72), 1·67 (1·61–1·73), and 1·69 (1·63–1·74), respectively; Re was not significantly different either among age cohorts or between waves. Conclusions: The pandemic generally caused higher mortality among young adults and might have spread mainly among school-aged children during the first wave. We propose mechanisms to explain the timing and transmission dynamics of the four NYC pandemic waves.