In this study, a series of large scale repeated model load tests are performed on geocell reinforced and unreinforced base layers overlying weak sand subgrades. The weak sand subgrades are prepared at 30% relative density (RD) through pluviation (sand raining) technique in a test tank of dimensions 1m × 1m × 1m (length × width × height). Two different base courses consisting of 75% RD sand and a granular base material have been tested. The 75% RD sand base course is also prepared by pluviation technique, while the granular base course is prepared in 5 layers, each of 50mm thick, by static compaction. A repetitive load of 0.97kN and 9.7kN was applied on the prepared base layer through a 150mm diameter plate to replicate the traffic load equivalent to a contact pressure of 550kPa. Loading was applied through a graphical user interfaced multi-purpose test software along with the help of a hydraulic power unit, hydraulic service manifold and sophisticated double acting linear dynamic 100kN capacity actuator which is connected to a 3.5m high, 200kN capacity reaction frame. Four different tests are conducted on both the base courses (75% RD sand and granular base) with and without reinforcement overlying the weak sand subgrade separately. There is a considerable amount of improvement observed for different number of cycles and plate settlements on quantification of traffic benefit ratios (TBR), cumulative plastic deformations (CPD) and rut depth reduction (RDR) for geocell reinforced base courses. However, geocell reinforced granular base course have shown a better improvement comparatively. © 2016 The Authors. Published by Elsevier B.V.