Ozone density profile over the Korean peninsula was obtained by UV radiometer onboard the KSR-III (Korea Sounding Rocket-III) launched in Nov. 2002 and compared with other observations from satellites and ozonesonde. Due to altitude limitations in this first test flight of the newly-developed system, the apogee of the rocket was still in the stratosphere. Most of the previous algorithm for the optical absorption technique assumed to have measurements out of the ozone layer but our situation provided an opportunity to consider an algorithm to retrieve the vertical profiles of O<sub>3</sub> number density for the rocket flight whose apogee is still in the ozone layer. In measurements by using optical instruments, various error sources exist in characterizing optical properties of detectors and atmospheric parameters. The magnitude of errors are analyzed and estimated in various rocket soundings but their quantitative effects of error sources due to each parameter in the retrieval have not been investigated in detail yet. In this paper, the quantitative error effects on the retrieval algorithm are investigated with respect to the altitudes. It is worthwhile to investigate this error sources for the current sounding as well as to reevaluate the previous rocket sounding data. Among retrieval parameters, the most influential error sources are found to be the absorption cross section and the filter response function. +0.65 nm shift in filter response function or -0.55 nm shift errors in absorption cross section are found to result in 5% deviation on the total number density. Whereas the total number density profiles are not so sensitive to the Rayleigh scattering or the slant air column density assumed. The stray light effect of the interference filter was also investigated.