The vibrational spectrum of Ar2HF in the 11 320-11 430 cm(-1) region is recorded by intracavity laser-induced fluorescence. The intramolecular vibrational state, Sigma(0), in combination with the intermolecular vibrations, assigned as Pi(in-plane), Pi(out-of-plane) and Sigma(1), of the complex have been observed. The Sigma(0) state correlates adiabatically with j = 0 of HF (upsilon = 3); the Pi(in-plane), Pi(out-of-plane), and Sigma(1) states correlate adiabatically with j = 1 of HF (upsilon = 3), respectively. We have determined the vibrational band origins (and rotational constants) of nu(0) = 11 323.784 cm(-1) (A = 0.120 15, B = 0.058 30, C = 0.038 94 cm(-1)), nu(0) = 11 387.730 cm(-1) (A = 0.122 68, B = 0.057 05, C = 0.038 42 cm(-1)), nu(0) = 11 426.815 cm(-1) (A = 0.120 27, B = 0.058 15, C = 0.038 71 cm(-1)) and nu(0) = 11 427.400 cm(-1) (A = 0.120 26, B = 0.058 15, C = 0.038 71 cm(-1)) for Sigma(0), Pi(in-plane), Pi(out-of-plane), and Sigma(1) states, respectively. The vibrational red shift for the pure HF stretch from upsilon(HF) = 0-3 is 49.023 cm(-1). The in-plane and out-of-plant bending frequencies are 63.947 and 103.031 cm(-1). The Sigma(1) state, which may be viewed as the Ar2FH structure is located 103.616 cm(-1) above the Ar2HF Sigma(0) state. The spectral line shapes appear to be well fitted by a Doppler profile with FWHM approximate to 120 MHz, indicating that the predissociation linewidths have a Lorentzian component of less than 10 MHz. These results are compared with those of Farrell and Nesbitt [J. Chem. Phys. 105, 9421 (1996)] for upsilon(HF) = 1. The present experimental data set is also compared with the quantitative predictions by Ernesti and Hutson [Phys. Rev. A 51, 239 (1995)] and therefore serves as a rigorous test for modeling nonadditivity of intermolecular interactions and their vibrational dependence. These comparisons show that the vibrational dependence of three-body terms is accurate in the region of potential minimum. For configurations far from the energy minimum, appreciable discrepancies appear to exist. The vibrational variation of the Pi(in-plane) bending frequency is relatively poorly predicted, which strongly suggests the inadequacy in the present modeling of the intriguing nonadditive forces for this prototypical system. (C) 1997 American Institute of Physics.