Abstract Two small multicellular convective areas within a larger mesoscale convective system that occurred on 20 June 2004 were examined to assess vertical motion, radar reflectivity, and dual-polarimetric signatures between flash and non-flash-producing convection. Both of the convective areas had similar life cycles and general structures. Yet, one case produced two flashes, one of which may have been a cloud-to-ground flash, while the other convective area produced no flashes. The non-lightning-producing case had a higher peak reflectivity up to 6 km. Hence, if a reflectivity-based threshold were used as a precursor to lightning, it would have yielded misleading results. The peak upward motion in the mixed-phase region for both cases was 8 m s−1 or less. However, the lightning-producing storm contained a wider region where the updraft exceeded 5 m s−1. Consistent with the broader updraft region, the lightning-producing case exhibited a distinct graupel signature over a broader region than the non-lightning-producing convection. Slight differences in vertical velocity affected the quantity of graupel present in the mixed-phase region, thereby providing the subtle differences in polarimetric signatures that were associated with lightning activity. If the results here are generally applicable, then graupel volume may be a better precursor to a lightning flash than radar reflectivity. With the dual-polarimetric upgrade to the national observing radar network, it should be possible to better distinguish between lightning- and non-lightning-producing areas in weak convective systems that pose a potential safety hazard to the public.