We have derived masses and radii for both components in five short-period single-lined eclipsing binary stars discovered by the TrES wide-angle photometric survey for transiting planets. All these systems consist of a visible F-star primary and an unseen M-star secondary (M A >= 0.8 M sun, M B <= 0.45 M sun). The spectroscopic orbital solution combined with a high-precision transit light curve for each system gives sufficient information to calculate the density of the primary star and the surface gravity of the secondary. The masses of the primary stars were obtained using stellar evolution models, which requires accurate determinations of metallicities and effective temperatures. In our case, the uncertainty in the metallicity of the primary stars is the most important limiting factor in order to obtain accurate results for the masses and radii of the unseen M-dwarf secondaries. The solutions were compared with results obtained by calculating the radius of the primary stars under the assumption of rotational synchronization with the orbital period and alignment between their spin axis and the axis of the orbit, using the observed broadening of the spectral lines as an indicator of stellar rotation. Four systems show an acceptable match between the two sets of results when their metallicity is allowed to vary around solar values (-0.5 <= [Fe/H] <= +0.5), but one system shows a clear mismatch between the two solutions, which may indicate the absence of synchronization or a misalignment between the rotational and orbital axis. When compared to low-mass stellar evolution models, the derived masses and radii of the unseen M dwarfs are inconsistent (three only marginally) with the predicted values, with all of the radii being larger than expected for their masses. These results confirm the discrepancy shown in a previous work between the predicted and observed radii on low-mass binary stars. This work also shows that reliance on the assumption of synchronization to derive the mass and radius of stars in eclipsing single-lined F+M binaries is a useful tool, but may not always be warranted and should be carefully tested against stellar evolution models.