Although resistive diffusion is much slower than the growth of resistive instabilities, the conventional neglect of diffusion in tearing mode calculations is incorrect. The proper criterion for neglect of diffusion in resistive instability calculations, which is not satisfied for tearing modes, is &ohgr;≫v/&dgr;, wherevis the resistive diffusion velocity and &dgr; is the resistive layer thickness. The effect of diffusion is calculated in the limit of large and small &ohgr;&dgr;/vfor the plane slab model, and new expressions for growth rate and stability boundary are obtained. The diffusion appears to have a stabilizing effect.