This thesis reports research on end anchorage at simple supports of reinforced concrete members and treats both straight bars and bars with 900 and 1800 bends. The most significant characteristics of straight anchorages at simple supports are their generally short lengths and the presence of transverse pressure from the support reactions. Published work in this area is rather limited. The only major research is that by Danish authors, working in the field of plasticity, and the only code of practice recommendations are those of Eurocode 2, which take account of the transverse pressure but do not consider the effects of the short lengths involved. Bends and hooks are widely treated in design codes, but their rules appear very arbitrary and seem to lack published substantiation. The approach adopted here is essentially empirical. A data base of results from tests of anchorages without transverse pressure is assembled and used to evaluate existing expressions for bond strength. An equation by Darwin, MaCabe, Idun and Schoenekase is found to be the most reliable of those considered and is modified in the light of the comparison. The most significant change is that the influence of the ratio of the anchorage length to the bar size is treated by a multiplying factor {! b I cp )04 , instead of being treated as an additional resistance independent of concrete strength, cover etc. In overall terms the modified equation produces a modest improvement in the correlation between calculated and actual strengths, but the above change and an alteration to the way in which covers and spacings are treated do improve reliability in areas which are important for end anchorages. Sixty five tests were made on end anchorages in simply supported beams. The bars had straight anchorages in thirty seven of the tests, 900 bends in fifteen and 1800 hooks in thirteen. The main variables were concrete cover, anchorage length, xxi transverse pressure and internal diameters of bends. The results of these tests, together with others from the literature are used to develop expressions for anchorage capacities. For straight ends the result is a bi-linear relationship between the ultimate bond stress and the transverse pressure (p). For p = 0 the bond resistance is that of the equation above and the gradient fbu / P is 2.0. For higher pressure fbu / P is 0.4. The correlation with the 186 test results is with the ratios between experimental and calculated strengths having a mean of 1.02 and a coefficient of variation of 14.7%. These figures compare favourably with the 1.93 and 17% for Ee2. For anchorages with terminal bends and hooks, the bar force developed bonded lead lengths over supports is calculated as for a straight bar with transverse pressure, and the bond strength in the bend+tail is that for a straight bar without transverse pressure. The bearing capacity of the lead is calculated as in BD44/95, which takes account of spread of stress away from the inside of the bend being three-rather than two dimensional. The total capacity of an anchorage is the sum of the forces developed by the lead length and the bend+tail, with the latter taken as the lesser of the values determined by bearing and bond. All lengths used in the calculations are the real dimensions and not effective lengths as used in some cases in BS811 0 and bearing stresses are checked in all cases. For the anchorage failures of bent and hooked bars, all but five of which are from the present tests the ratios of experimental to calculated strengths have a mean of 1.10 and a coefficient of variation of 15 % which compare with values of 1.60 and 17 % for BS811 0, 1.40 and 18 % for Ee2 and 1.59 and 17% for BD44/95. The number and range of the test results is too limited to properly confirm the reliability of the present approach, but it does appear to the considerably more reliable than current design methods and avoids the use of fictitious lengths and arbitrary omissions of checks on bearing stresses.