Passive acoustic data are increasingly being used as a tool for helping to define marine mammal populations and stocks. Fin whale (Balaenoptera physalus) songs present a unique opportunity to determine interstock differences. Their highly stereotyped interpulse interval has been shown to vary between geographic areas and to remain stable over time in some areas. In this study the structure of songs recorded at two geographically close feeding aggregations in the Gulf of St. Lawrence (GSL) and Gulf of Maine (GoM) was compared. Recordings were made from September 2005 through February 2006 in the GSL and intermittently between January 2006 and September 2007 at two locations in the GoM. 6257 pulse intervals corresponding to 19 GSL and 29 GoM songs were measured to characterize songs from both areas. Classification trees showed that GSL songs differ significantly from those in the GoM. The results are consistent with those derived from other stock structure assessment methodologies, such as chemical signature and photoidentification analysis, suggesting that fin whales in these areas may form separate management stocks. Song structure analysis could therefore provide a useful and cost-efficient tool for defining conservation units over temporal and geographical scales relevant to management objectives in fin whales.