We describe a new method of calibrating tropospheric hydroxyl (OH) instruments. Ozone-alkene mixtures produce steady-state OH radical concentrations. The steady state is governed by competition between OH production in the reaction of ozone with the alkene and OH removal by reactions with the alkene, ozone, and the reactor wall. In a flowtube reactor transporting an ozone-alkene mixture, the OH wall loss rate can be measured by varying the alkene concentration. Delivery of the reaction mixture to the sampling entry of an atmospheric OH measurement instrument provides an absolute calibration of the instrument's response to OH. The present precision of calibration is +/-8% (1-sigma), based on reproducibility over a wide range of ozone concentrations. The accuracy (+/-43%) is limited by uncertainties in kinetic rate coefficients and OH yield, which can be improved. The calibration requires no photon flux measurements or lamp-dependent absorption coefficients, which have inherent, variable, systematic uncertainties, and it has been tested in field experiments.