A magnitude limited population of 18808 Hipparcos stars is used to calibrate distances for 52794 RAVE stars, including dwarfs, giants, and pre-main sequence stars. I give treatments for a number of types of bias affecting calculation, including bias from the non-linear relationship between the quantity of interest (e.g., distance or distance modulus) and the measured quantity (parallax or visual magnitude), the Lutz-Kelker bias, and bias due to variation in density of the stellar population. The use of a magnitude bound minimises the Malmquist and the Lutz-Kelker bias, and avoids a measurement bias because Hipparcos parallaxes are more accurate for brighter stars. The calibration is applicable to stars in 2MASS when there is some way to determine stellar class with reasonable confidence. For RAVE this is possible for hot dwarfs and using log g. The accuracy of the calibration is tested against Hipparcos stars with better than 2% parallax errors, and by comparison of the RAVE velocity distribution with that of Hipparcos, and is found to improve upon previous estimates of luminosity distance. An estimate of the LSR from RAVE data, (U0, V0, W0) = (14.9-1.7, 15.3-0.4, 6.9-0.1)km/s shows excellent agreement with the current best estimate from XHIP. The RAVE velocity distribution confirms the alignment of stellar motions with spiral structure.