The land vehicle navigation using very few GPS updates is studied in this paper using a low-cost MEMS sensors assembly integrated with a conventional GPS receiver. The sensors assembly is composed of triad of accelerometers, gyrometers, magnetometers as well as a pressure sensor, and can easily be taken in or off the car. Since in most of cities, downtown areas block GPS signals, conventional receivers may not be able to compute 3D or even 2D fixes. Therefore, to augment the availability of the position solution in harsh environments, pseudorange and pseudorange rate measurements are fused in a Kalman Filter (KF) with the Dead Reckoning (DR) solution provided by the sensors assembly following a tight integration scheme. In the developed vehicular navigation algorithm, the MEMS sensors are used to provide reliable attitude information of the vehicle. It also propagates the position of the vehicle in three dimensions. A conventional GPS receiver gives pseudorange and pseudorange rate measurements for the tight integration. External ephemeredes are used to estimate the satellite Doppler contribution in the user-to-satellite Line Of Sight (LOS). Results show that as soon as two satellites are tracked, the overall accuracy of the integrated positioning system increases. Moreover, the availability of the position solutions is also augmented. Depending on the type of harsh environments, namely classical or deep urban canyons, heavy map-matching algorithms are not required with such low-cost sensors integration. However, the integration scheme is dependent on the satellite geometry with respect to the user's heading. When a bad satellite configuration is detected, the Integrated Navigation System (INS) relies exclusively on the IMU data. As a consequence, the INS may experience systematic drift for long periods (>1min) without any GPS updates.