We investigate the residual stress in diamond films grown on ͑001͒ silicon substrates as a function of film thickness. The diamond films were deposited at 1070 K by the conventional hot filament technique using a gas mixture of methane ͑1.0% vol͒ and hydrogen ͑99.0% vol͒. The film thickness, obtained from cross section scanning electron micrographs, varied from 3.0 to 42 m as the growth time increased from 1 to 10 h. These images evidenced that the columnar growth is already established for films thicker than 10 m. Top view micrographs revealed predominantly faceted pyramidal grains for the films at all growth stages. The grain size, obtained from these images, was found to vary linearly with film thickness. Using a high resolution x-ray diffractometer, the residual stress was determined by measuring, for each sample, the ͑331͒ diamond Bragg diffraction peak for ⌿ values ranging from Ϫ60°to ϩ60°, and applying the sin 2 method. For the micro-Raman spectroscopy, we used the summation method, which consists in recording and adding a large number of spectra in different places of a selected area of the sample. All Raman spectra were fitted with Lorentzian lines to separate the contribution of the pure diamond and the other nondiamond ͑graphite͒ phases. This spectral analysis performed in each sample allowed the determination of the residual stress, from the diamond Raman peak shifts, and also the diamond purity, which increases from 70% to 90% as the thickness goes from 3 to 42 m. The type and magnitude of the residual stress obtained from x-ray and micro-Raman measurements agreed well for films thicker than 10 m. For films thinner than this value, an opposite behavior between both results was observed. We attributed this discrepancy to the domain size characteristic of each technique.