Large databases of cyclic tests on flexure‐ or shear‐critical concrete members and shear‐critical connections of columns to beams or slabs are used to estimate the uncertainty inherent in experimental data in literature—as read by users. To this end, the predictions of two different, presumably independent, design‐oriented models for the properties of interest are used to establish the “central tendency” of data, against which individual tests or small groups thereof are assessed. Properties considered are: (a) the cyclic ultimate chord‐rotation of flexure‐controlled members with continuous or lap‐spliced deformed bars, (b) the cyclic shear strength of shear‐critical members, (c) the chord‐rotation at yielding of rectangular columns with plain bars, and (d) the cyclic shear strength of shear‐controlled beam‐column and slab‐column joints. Results suggest that the data from each test campaign have a certain degree of bias, specific to it. Test campaigns with ratio of estimated average deviation from the “central tendency” to the standard deviation of campaign deviations (called “data uncertainty”) which is far into the tail of the Normal distribution may be excluded as questionable. This systematic bias, along with other types of “data uncertainty” addressed in this work, seem to contribute to the apparent scatter of model predictions with respect to cyclic test results the equivalent of a coefficient of variation of model‐to‐test‐ratio of at least 10% and possibly as high as 25%–30%. Model uncertainty seems to contribute to this scatter the equivalent of a coefficient of variation of at least 15% in shear‐controlled connections, or as much as 25% in the case of flexural deformation capacity of members with deformed bars; the cyclic shear resistance of members and—with the reservation of the small number of tests—the chord‐rotation at yielding of members with plain bars, are in‐between.