Today rapid-static GPS is an alternative surveying technique among other fundamental GPS surveying methods such as real-time kinematic and static GPS. It usually finds application areas in engineering surveying and monitoring of ground deformation for example landslides. Web-based GPS software such as OPUS-RS which was developed by the National Geodetic Survey (NGS) promoted further the use of the technique. NGS also provides its users the obtainable accuracy of the OPUS-RS derived solutions. Minimum 15 min of GPS data is recommended for rapid-static positioning and the accuracy is given accordingly. In this study, we go beyond the typically recommended 15 min observations and examine the accuracy variation of rapid-static OPUS positioning solutions over 8-118 min interval. Seven Continuously Operating Reference Stations from the US are selected, and their data are segmented into the above-mentioned shorter sessions. Statistical analyses of the CORS stations revealed that solutions from 15 min or shorter sessions contain outliers, and this degrades the efficiency of the technique. By extending the typical 15 min observation length up to 60 min perfectly eliminates the outliers, and the accuracy on the average is improved by about 5-20 mm for horizontal and about 25 mm for vertical components.