Recently, fMRI was used to estimate visual population receptive fields (pRFs) by fitting a model shape to the measurements, permitting estimates of visual field coordinates, pRF size, and other parameters (Dumoulin & Wandell, Neuroimage 39:647, 2008). Here we demonstrate that pRFs can be directly reconstructed using a tomographic approach that requires no a priori shape assumption. Methods: We obtained fMRI images across early visual cortex using a GE 3T scanner using 2-mm voxels and a spiral acquisition. Stimulation was a thin bar (1–2.5° wide) containing a moving checkerboard pattern that swept slowly (<1°/s) across a 10–20° field-of-view. Bar motion was perpendicular to its long axis. The motion direction and bar orientation of each sweep successively rotated from 0–165° in 15° increments. Blank periods were inserted between each sweep to allow the fMRI response to subside. Assuming linearity, the fMRI time series corresponded to a projection of the pRF along the axis of the bar. The multiple sweeps thus formed a sinogram (Radon transform) that was blurred by the hemodynamic response function (HRF). Blurring was mitigated using a Wiener filter incorporating HRF waveforms measured in each scanning session. PRFs were then reconstructed from the corrected sinograms using a backprojection algorithm. Contours were created around the pRF half-maxima to estimate field coordinates and other parameters. Results: Our polar angle and eccentricity maps agree well with the previous model-based approach. PRF sizes increase with eccentricity and with visual area. Notably, individual reconstructed pRFs are complex with anisotropic (non-circular) shapes and significant suppressive surround regions. Anisotropy and suppression exhibit trends with respect to visual area and eccentricity. We present a detailed analysis of the tomographic spatial resolution that is essential to interpretation of these trends. Conclusion: Tomographic reconstruction of pRFs is a useful approach for estimating visual receptive field properties without a priori shape assumptions.