Accurate measurement of the true vertical depth (TVD) and reduction in the error uncertainty of the TVD measurement has become an increasingly important element in field development and well placement. A field on the North West Shelf of Australia required precise positioning of the wellbore within the reservoir. The trajectory within the horizontal production section depended completely on accurate geometric measurements to achieve the goal of keeping the well within +/-1 m of the target TVD for the entire length of the horizontal section – typically 1600m to 2300m. With great attention to detail, this is possible; however, the uncertainties associated with determining the actual wellbore inclination would normally overwhelm the required accuracy. The main sources of error while drilling horizontally result from misalignment (SAG) of the bottomhole assembly (BHA) within the wellbore, sensor axial misalignment, and axial accelerometer bias. To combat these sources of error, a multiple sensor system was designed with improved accelerometer bias (attained by using extreme calibration techniques). Compounded software programs for advanced BHA analysis and multistation analysis of the survey data were used. The data from the primary sensor determined the directional parameters used while drilling. The second directional sensor was positioned in the BHA to confirm the accuracy of the SAG correction applied to the primary sensor and to confirm any bias. Proprietary software was used in real time and post-drilling to perform multistation analyses of the accelerometer and magnetometer data from both sensors. This paper describes the methods used to reduce the TVD uncertainty while drilling long horizontal sections. Improvements were made to standard Industry Steering Committee for Wellbore Surveying Accuracy (ISCWSA) MWD+SAG error model to simulate the increased level of accuracy generated as a result of the multiple sensor system. Using data from the two directional sensors and a continuous at-bit accelerometer assembly, the ellipse of uncertainty was estimated throughout the section and remained with a +/-1.5 m (1 sigma) limit. TVD uncertainty was set to zero at the gas/oil contact (GOC); it was not necessary to model the total error from surface.