Clouds can significantly affect photochemical activities in the boundary layer by altering radiation intensity, and therefore their correct specification in the air quality models is of outmost importance. In this study we introduce a technique for using the satellite observed clouds to correct photolysis rates in photochemical models. This technique was implemented in EPA's Community Multiscale Air Quality modeling system (CMAQ) and was tested over a 10 day period in August 2000 that coincided with the Texas Air Quality Study (TexAQS). The simulations were performed at 4 and 12 km grid size domains over Texas, extending east to Mississippi, for the period of 24 to 31 August 2000. The results clearly indicate that inaccurate cloud prediction in the model can significantly alter the predicted atmospheric chemical composition within the boundary layer and exaggerate or underpredict ozone concentration. Cloud impact is acute and more pronounced over the emission source regions and can lead to large errors in the model predictions of ozone and its by-products. At some locations the errors in ozone concentration reached as high as 60 ppb which was mostly corrected by the use of our technique. Clouds also increased the lifetime of ozone precursors leading to their transport out of the source regions and causing further ozone production down-wind. Longer lifetime for nitrogen oxides (NOx = NO + NO2) and its transport over regions high in biogenic hydrocarbon emissions (in the eastern part of the domain) led to increased ozone production that was missing in the control simulation. Over Houston-Galveston Bay area, the presence of clouds altered the chemical composition of the atmosphere and reduced the net surface removal of reactive nitrogen compounds. Use of satellite observed clouds significantly improved model predictions in areas impacted by clouds. Errors arising from an inconsistency in the cloud fields can impact the performance of photochemical models used for case studies as well as for air quality forecasting. Air quality forecast models often use the model results from the previous forecast (or some adjusted form of it) for the initialization of the new forecast. Therefore such errors can propagate into the future forecasts, and the use of observed clouds in the preparation of initial concentrations for air quality forecasting could be beneficial.