The accuracy of the counterflow, twin-flame technique for the determination of laminar flame speeds was examined analytically, numerically and experimentally. The analysis was conducted by using multiple-expansion, large activation energy asymptotics, while the numerical simulation incorporated detailed chemistry and transport. In both approaches the solutions were obtained in a finite domain and with plug flow boundary conditions in order to better simulate the actual experiments. Results show that linear extrapolation of the minimum velocity to zero stretch overestimates the true laminar flame speed. This overestimate, however, can be reduced by using larger nozzle separation distances. The theoretical results were further confirmed by experimental measurements for methane/air flames with various stoichiometries and nozzle separation distances. The numerical and experimental results indicate that for atmospheric methane/air flames, nozzle separation distances in excess of about 2 cm yield laminar flame speeds obtained by linear extrapolation accurate to within the uncertainty range of the experiment. The results obtained herein thus provide further support for the viability of the counterflow technique, when the influence of the nozzle separation distance is properly accounted for. The viability of an alternate technique for the determination of laminar flame speeds, based on the variation of how velocity at a constant temperature near the upstream boundary of the flame with stretch, was also theoretically investigated.