We combine results from interferometry, asteroseismology and spectroscopy to determine accurate fundamental parameters of 23 bright solar-type stars, from spectral type F5 to K2 and luminosity classes III-V. For some stars we can use direct techniques to determine the mass, radius, luminosity and effective temperature, and we compare with indirect methods that rely on photometric calibrations or spectroscopic analyses. We use the asteroseismic information available in the literature to infer an indirect mass with an accuracy of 4-15 per cent. From indirect methods we determine luminosity and radius to 3 per cent. We find evidence that the luminosity from the indirect method is slightly overestimated (≈ 5 per cent) for the coolest stars, indicating that their bolometric corrections (BCs) are too negative. For Teff we find a slight offset of -40 ± 20 K between the spectroscopic method and the direct method, meaning the spectroscopic temperatures are too high. From the spectroscopic analysis we determine the detailed chemical composition for 13 elements, including Li, C and O. The metallicity ranges from [Fe/H]=-1.7 to +0.4, and there is clear evidence for α-element enhancement in the metal-poor stars. We find no significant offset between the spectroscopic surface gravity and the value from combining asteroseismology with radius estimates. From the spectroscopy we also determine v sin i and we present a new calibration of macroturbulence and microturbulence. From the comparison between the results from the direct and spectroscopic methods we claim that we can determine Teff, log g and [Fe/H] with absolute accuracies of 80 K, 0.08 and 0.07 dex. Photometric calibrations of Strömgren indices provide accurate results for Teff and [Fe/H] but will be more uncertain for distant stars when interstellar reddening becomes important. The indirect methods are important to obtain reliable estimates of the fundamental parameters of relatively faint stars when interferometry cannot be used. This paper is the first to compare direct and indirect methods for a large sample of stars, and we conclude that indirect methods are valid, although slight corrections may be needed.