Determination of the strike of the dominant 2D geoelectrical structure from magnetotelluric (MT) data is difficult in the presence of both noise and local distortion. This paper illustrates that often the strike angle is the least stable parameter which can be resolved from the MT data themselves, and that the telluric distortion parameters are usually more robustly estimated. In particular, techniques which rely only on the analytical rotational properties of the MT impedance tensor can yield erroneous results in the presence of even small amounts of noise. The estimation procedure is stabilized once the telluric distortion parameters have been correctly determined and a constrained fit is sought. Also, it is emphasized that even when the true strike angle is known, statisticaly superior estimates of the regional impedances result when fitting a galvanic distortion model to the data rather than merely rotating the impedance tensor into the determined (or assumed) strike direction.