Measurements of dose distribution for square fields with sizes ranging from 1 X 1 to 30 X 30 cm for a 9-MV x-ray beam from a Neptune 10 linear accelerator, manufactured by CGR, are reported. Special attention was paid to field sizes smaller than 4 X 4 cm, used in radiosurgery techniques. To express the dose-monitor units relationship, total, collimator, and phantom scatter correction factors were obtained by experimental measurements. A strong dependence of these factors on the smallest field sizes (less than 4 X 4 cm) was shown. Measurements of the maximum depth dose dmax, plotted as a function of field size, showed a maximum at about 5 X 5 cm, in good agreement with previous results. dmax was also measured for the smallest fields, demonstrating that the contaminating electron component of the x-ray beam was not responsible for the dmax shift. Analysis of the penumbra width of cross dose distributions, as a function of field sizes, allowed us to postulate that the dmax shift could be due to the phantom scattered photons, which in turn were generated by the collimator scattered photons. Newly derived tissue-maximum ratio and scatter-maximum ratio data were used for dose profile calculations of 2 X 2, 4 X 4, and 10 X 10 cm field sizes. The agreement between experimental and calculated data was found to be +/- 2% within the geometrical edges of the fields and +/- 6% outside of them. A dose profile from the isocenter of a 2 X 2 cm square field moving through a 360 degree rotation arc was obtained and compared with that from the center of a 125I shielded source, as measured by Ling. Advantages and problems relating to the use of x-ray beams from linear accelerators in radiosurgery are discussed.