Includes bibliographical references. ; Deflection under service loading is an important aspect of reinforced concrete slab design. Under-design can cause large deflections which can be expensive to repair, if at all possible. Over-design can lead to material wastage and unnecessary dead load. Deflection is inversely proportional to the effective moment of inertia of the section under consideration. Cracks, which may or may not be present at the serviceability limit state, have a profound effect on the moment of inertia. Many Codes of practice approach the calculation of deflection in a conservative manner by using the cracked moment of inertia in deflection calculations and ignoring the effect of the concrete in tension. Two of the Codes reviewed make an attempt at including the stiffening effect of the concrete in tension. The theory in the CEB/FIP Model Code is used as a basis for the method that is developed to predict maximum deflections. This method proposes that the total maximum deflection is composed of two components: an elastic deflection and a deflection due to cracking. The elastic deflection for a beam is determined from elastic formulae that are developed from first principles for standard beam cases. The deflection due to cracking involves the cracking moment capacity of the beam, what portion of the beam is cracked, the formation of a hinge and the rotation of this hinge. One-way spanning slabs can be treated as broad, shallow-beams. Two-way spanning slabs are more complicated and to determine the load dispersion of a uniformly distributed load on such a slab, it is divided into five sets of orthogonal strips. The two outer strips do not carry any load. The three inner strips intersect at nine points or nodes. The deflection of each pair of orthogonal strips at each of the nine nodes must be equal. Deflection equations are set up in terms of an unknown portion of the load at each node. Since the full load at each node is known, the sum of the loads in the orthogonal directions must be equal to this full load. A matrix is set up and solved and the load dispersion at each node is determined. The equivalent load on a strip spanning through the region of maximum deflection is thus found. For the two orthogonal strips spanning through the region of maximum deflection, the average deflection is then taken. A computer program is written which incorporates the above approach. The program is then run for slab configurations that were tested in the laboratory and the results are compared. The results show that the proposed computational models over-predict slab deflections. As soon as the slab is clamped on more than one edge or if the aspect ratio increases above 1 then the results in the orthogonal directions differ by a large amount. The recommended improvements to these computational models are: - Increase the number of orthogonal strips and introduce torsion. This will also improve the continuity between strips spanning in the same direction. The tension stiffening factor has to be redefined. This will reduce the contribution of deflection due to cracking.