We investigate the relationship between spiral arms and star formation in the grand-design spirals NGC 5194 and NGC 628 and in the flocculent spiral NGC 6946. Filtered maps of near-IR (3.6 micron) emission allow us to identify "arm regions" that should correspond to regions of stellar mass density enhancements. The two grand-design spirals show a clear two-armed structure, while NGC 6946 is more complex. We examine these arm and interarm regions, looking at maps that trace recent star formation - far-ultraviolet (GALEX NGS) and 24 micron emission (Spitzer, SINGS) - and cold gas - CO (Heracles) and HI (Things). We find the star formation tracers and CO more concentrated in the spiral arms than the stellar 3.6 micron flux. If we define the spiral arms as the 25% highest pixels in the filtered 3.6 micron images, we find that the majority (60%) of star formation tracers occurs in the interarm regions; this result persists qualitatively even when considering the potential impact of finite data resolution and diffuse interarm 24 micron emission. Even with a generous definition of the arms (45% highest pixels), interarm regions still contribute at least 30% to the integrated star formation rate tracers. We look for evidence that spiral arms trigger star or cloud formation using the ratios of star formation rate (SFR, traced by a combination of FUV and 24 micron emission) to H_2 (traced by CO) and H_2 to HI. Any enhancement of SFR / M(H_2) in the arm region is very small (less than 10%) and the grand design spirals show no enhancement compared to the flocculent target. Arm regions do show a weak enhancement in H_2/HI compared to the interarm regions, but at a fixed gas surface density there is little clear enhancement in the H_2/HI ratio in the arm regions. Thus, it seems that spiral arms may only act to concentrate the gas to higher densities in the arms.