Extratropical cyclones are responsible for significant weather in the form of heavy precipitation and strong winds. The capability of numerical weather prediction models to predict the synoptic-scale structure of such cyclones has improved greatly over recent years but much of the significant weather itself is associated with small and mesoscale processes not properly represented even in today's relatively high-resolution models. As a result, the detailed prediction of significant weather, even for the period 1–12 h ahead, still falls far short of requirements. In order to find out what improvements are needed by way of increased model resolution, better parameterizations, and/or improved observations/assimilation, it is first necessary to learn more about the structure, mechanism, and interaction of the small-scale and mesoscale processes. This is the subject of this review. The review focuses on the structure and organization of slantwise and upright convection within extratropical cyclones, particularly cold-season maritime cyclones. These subsynoptic-scale features are set within a broader context using the conveyor-belt and frontal-fracture paradigms. It is shown that there is a common tendency for slantwise convection to occur in the form of vertically stacked multiple circulations, sometimes associated with lines of upright convection that are themselves broken into chains of line elements. The review also examines the nature and significance of evaporation/sublimation and shearing instability within frontal zones. These processes play opposing roles in, respectively, sharpening and diffusing the individual slantwise convective circulations. Although shearing instability occurs mostly at very small scales, individual events can lead to breakdown of laminar flow over layers as much as 1 km deep. Such events may be attributed to potential shearing instability, which, like its counterpart in convective instability, can suddenly be released where a layer of air is lifted to saturation. The studies of the above processes described in this review make extensive use of observations from many different types of radar. Although it is generally necessary to interpret radar observations within the context of other information, the pivotal role of radar in these studies is clear. The writer owes a debt of gratitude of Dave Atlas who from an early stage inspired him to attempt to use the full potential of radar.