In a method for forming a direct positive color image which comprises image-wise exposing to light a light-sensitive material comprising a support having provided thereon at least one photographic emulsion layer contaiing unfogged internal latent image type silver halide grains and a color image-forming coupler; developing the light-sensitive material with a surface developing solution containing an aromatic primary amine type color developing agent in the presence of a nucleating agent and/or fogging exposure; bleaching and fixing the resulting material to form the direct positive color image, the improvement is characterized in that the developing solution has a pH of 11.5 or less and does not substantially contain benzyl alcohol; the color coupler is a compound which forms or releases a substantially non-diffusive dye by oxidation coupling with a developing agent, and the internal latent image type silver halide grains have a particular core/shell structure.According to the method, a direct positive color image having a high maximum color density and a low minimum image density can be obtained in a short time.