We present two new experimental methods of realizing the desired kinematics in a filament stretching rheometer. Due to the presence of the rigid plates connecting the fluid filament in this device, a homogeneous uniaxial elongational flow cannot be imposed throughout the entire fluid filament. However, if the fluid element at the midpoint of the filament is forced to contract in the same manner as a cylindrical column undergoing ideal uniaxial extension, then the measured rheological response will be virtually identical to that experienced in a homogeneous shearfree flow. We investigate both a real-time active control scheme and a second, original technique based on a one-dimensional slender-filament approximation of the kinematics as possible methods of realizing this form of nearly-ideal filament stretching experiment. We find that the real-time active control scheme induces undesirable oscillations in the imposed plate separation profile, leading to unacceptable fluctuations in the measured force data and limiting the achievable operating range of the device. The second technique, on the other hand, produces very accurate and smooth mid-point diameter profiles in a two-step process. Transient Trouton ratio data obtained by using the second method with both a polystyrene-based Boger fluid and a polyisobutylene-based Boger fluid agree well with data previously reported in the literature. Differences between the transient Trouton ratios at intermediate strains obtained from experiments in which the midpoint diameter decreases either ideally or inhomogeneously are consistent with expectations from recent simulations. The second technique presented here provides a simple method of achieving accurate filament stretching data that can be compared quantitatively with theoretical constitutive models. (C)1999 Elsevier Science B.V. All rights reserved.