An analysis is presented of the large-scale conditions associated with the initial development of Tropical Storm Diana (September 1984) in a baroclinic environment. Ordinary extratropical wave cyclogenesis began along an old frontal boundary east of Florida after 0000 UTC 7 September and culminated in tropical cyclogenesis 48 h later. Water-vapor satellite imagery showed that the initial cyclogenesis and incipient tropical storm formation was nearly indistinguishable from a classical midlatitude development. Cyclogenesis occurred in three stages. A large-scale cold trough and associated frontal system crossed the Atlantic coast, while a small potential vorticity maximum aloft fractured off the main trough and stalled over central Florida in the first stage. As the main trough sheared off eastward, cyclogenesis began along the southwestern end of the stalled frontal zone east of Florida. Anticyclogenesis to the north in the wake of the shearing trough allowed a surge of cooler and drier air to flow southeastward behind the front toward the developing cyclone. Combined surface sensible and latent heat fluxes in excess of 1000 W m−2 acted on this inflowing air, producing a warming and moistening of the boundary layer. Cyclogenesis intensified during the second stage in response to positive potential vorticity advection aloft ahead of the slow moving cutoff cyclone over Florida. The maximum ascent was centered near 300 mb, indicative of deep tropospheric ascent and cyclonic vorticity production by convergence in midlevels. The ascent occurred along uplifted isentropic surfaces that defined the cold dome associated with the potential vorticity anomaly aloft. Low-level potential vorticity was generated in the vicinity of the developing storm below the presumed level of maximum diabatic heating. The third stage of cyclogenesis was marked by the collapse of the mid- and upper-tropospheric cold dome and associated potential vorticity maximum and the simultaneous initiation of a warm thickness ridge. This occurred in response to the widespread outbreak of convection at the southwestern end of the baroclinic zone, where the greatest destabilization occurred for air parcels subject to prolonged surface sensible and latent heat fluxes in the persistent northeasterly flow. Upright ascent associated with the convection short-circuited the slantwise ascent ahead of the advancing potential vorticity anomaly, triggering warming aloft and the eventual disappearance of the potential vorticity anomaly and associated cold dome. Tropical storm development and intensification occurred as the low-level vorticity center (potential vorticity maximum) moved northwestward to become situated beneath the midlevel vortex embedded within a local 500–200 mb warm thickness anomaly. The interaction of the upper- and lower-level potential vorticity anomalies appeared to be important in the initial strengthening of the tropical cyclone. The interpretation is equivalent to earlier energetic arguments by Riehl and others that tropical cyclogenesis is often preceded by the collapse of a nearby cold dome.