In the process of detecting thin film thickness with reflectometric interference spectroscopy (Rifs) methods, a phenomenon of "double peak" often appears in reflected spectrum, which seems like some adjacent peaks being connected very closely, and which is different from reflected spectrum as usual. Spectrum analysis and simulating are used to find why this occurs. It is found that this phenomenon is associated with the uneven surface of film, when uneven degree of film increases, "double peak" begin to appear. That is, when thickness value of detecting point changing, light was reflected at different thickness point of film, all these reflected spectrum are added to produce total reflected spectrum, and lead "double peak" to occur. "Double peaks" are easy to see when film thickness changes greatly. But simulating results also show the "double peak" won't affect detecting precision, because the wavelength position of extrema being same as usual position of reflected spectrum without "double peak". The reason is that the wavelength position of extrema is the only important thing to calculate film thickness, and the thickness being calculated through Rifs is the mean value of film. A series of simulating results are provided, as well as actual experimental result with "double peak" phenomenon. Simulating results of polymerized film show that the "double peak" occurs after thickness difference exceed 1/10. Other reasons that may lead to "double peak" phenomenon are also discussed in the paper.