Ubiquitous positioning using combined lowcost, Micro-electro-mechanical-systems (MEMS) Inertial Navigation Systems (INS) and Global Navigation Satellite System (GNSS) systems remains a challenge in difficult GNSS environments. This is due to the GNSS signals being partially or completely obstructed by building structures and consequently no information being available to calibrate the MEMS INS errors. In these environments although there may be insufficient signals for a two or three dimensional GNSS solution, some signals can still be received, particularly when utilizing high sensitivity GNSS receivers. These signals can be used to aid the INS/GNSS integrated system but can only be realized when it utilizes a tightly coupled (TC) as opposed to a loosely coupled (LC) integration architecture. This is particularly useful for MEMS based INS/GNSS integrated system as its performance can degrade rapidly over short period of GNSS outages. This paper aims to evaluate the capabilities of MEMS based INS/GNSS using TC integration architecture during extended period of GNSS outages. The results obtained show that TC integration architecture does help limit the error growth even when INS/GNSS integrated system experienced relatively long period of partial GNSS outages. commercially available MEMS INSs. Two independent datasets were collected and used to demonstrate the performance of the integrated system architecture (LC vs. TC) during periods of complete and partial GNSS outage. The results of these experiments conducted show that considerable improvements are observed when TC integration architecture is employed compared to LC. The experiment platform, integrated processing architecture and results obtained will be fully presented in this paper.