This thesis reports an investigation on time-dependent loss of post-tensioned masonry box and tee sections representing diaphragm and fin walls, respectively. The prestress loss due to creep and shrinkage of masonry, and relaxation of steel bars were quantified separately and the main influencing factors considered were geometry and masonry type. For each type of masonry three diaphragm and three fin walls were built to determine prestress loss (decreasing load), creep (constant load) and shrinkage (zero load). The walls were constructed from undocked clay, calcium silicate and concrete block units with grade (ii) mortar with cementlime:sand in the proportions of 1:½:4½ and water/cement ratio of 1.27. Creep and shrinkage were also measured on unbonded masonry units and mortar prisms for predicting the deformations in the masonry walls by using a previously developed composite model. The mal;onry units and mortar prisms were partly sealed to simulate the corresponding volume/surface ratio of the bonded masonry units and mortar joints in the masonry walls. The calcium silicate walls exhibited the highest prestress loss, creep and shrinkage compared with the clay and concrete block diaphragm and fin walls. The current methods of prediction of prestress loss for masonry are only suitable for specific types of masonry for which they were developed. On the other hand, the methods developed for prestressed concrete gave reasonable predictions for all the masonry types investigated, with one particular method being very accurate. For all test results it was confirmed that long-term deformations were influenced by geometry, fin walls exhibiting greater deformations than diaphragm walls. The composite model did not predict shrinkage very well in calcium silicate and concretc block walls because some moisture in the mortar was absorbed by the masonry units. As a result the partly sealed unbonded mortar prisms had higher water content than the mortar bed joint in the walls, and thus a higher shrinkage in the partly sealed mortar prisms occured. Consequently, when the creep and shrinkage of the partly sealed mortar prisms was applied to the model, the masonry deformation was overestimated. A modified water absorption test was carried out which confirmed that for units laid dry the mortar bed joint had a reduced shrinkage compared to the unbonded mortar prisms. From the results, creep and shrinkage adjustment factors were correlated with unit water absorption, and when adjusted creep and shrinkage were incorporated with the composite model, satisfactory predictions of masonry deformations were achieved.