We present a comparison of parallaxes and radii from asteroseismology and Gaia DR1 (TGAS) for 2200 Kepler stars spanning from the main sequence to the red-giant branch. We show that previously identified offsets between TGAS parallaxes and distances derived from asteroseismology and eclipsing binaries have likely been overestimated for parallaxes mas ( 90%-98% of the TGAS sample). The observed differences in our sample can furthermore be partially compensated by adopting a hotter scale (such as the infrared flux method) instead of spectroscopic temperatures for dwarfs and subgiants. Residual systematic differences are at the 2% level in parallax across three orders of magnitude. We use TGAS parallaxes to empirically demonstrate that asteroseismic radii are accurate to 5% or better for stars between . We find no significant offset for main-sequence ( ) and low-luminosity RGB stars ( 3-8 ), but seismic radii appear to be systematically underestimated by 5% for subgiants ( 1.5-3 ). We find no systematic errors as a function of metallicity between to dex, and show tentative evidence that corrections to the scaling relation for the large frequency separation ( ) improve the agreement with TGAS for RGB stars. Finally, we demonstrate that beyond asteroseismology will provide more precise distances than end-of-mission Gaia data, highlighting the synergy and complementary nature of Gaia and asteroseismology for studying galactic stellar populations.