Femtosecond stimulated Raman spectroscopy (FSRS) has emerged as a powerful new technique that is capable of obtaining resonance Raman spectra of fluorescent species and transient photochemical intermediates. Unlike related transient infrared absorption techniques, the FSRS signal is quite sensitive to the laser power utilized in the vibrational probing event. In particular, FSRS spectra are highly sensitive to the intensity of the picosecond Raman-pump pulse. We have measured the power dependence of the FSRS signal using pulse energies from ~10−9 to ~10−5 J and molecules with a range of molar absorptivities at the Raman-pump wavelength of 400 nm, including β-carotene (ε400 = 58 300 M−1 cm−1), para-nitroaniline (17 800 M−1 cm−1), nitronaphthalene (247 M−1 cm−1) and ferrocene (57 M−1 cm−1). We show that for strongly absorbing molecular systems, such as β-carotene and para-nitroaniline, the ground-state (GS) FSRS signal actually decreases with increasing pump power at pump fluences above ~10−2 J cm−2, due to depletion of the GS population. However, for weakly absorbing species like nitronaphthalene and ferrocene, the signal increases linearly with increasing pump fluence until ~0.5 J cm−2, at which point two-photon absorption by the solute induces nonlinear absorption of the pump pulse and attenuation of the FSRS signal. The data are quantitatively simulated with a photophysical kinetic model, and the results are analyzed to provide simple guidelines for acceptable Raman-pump powers in resonance FSRS experiments. The acceptable Raman-pump power is proportional to the focused beam area and depends inversely on the sample's molar absorptivity. Copyright © 2013 John Wiley & Sons, Ltd.