This paper presents the latest results of a joint research project involving the Institute of Microengineering of the Swiss Federal Institute of Technology (Switzerland) and the Industrial Automation Laboratory of the Lyon National Institute of Applied Sciences (France). This project consists in the implementation of a robotic assembly method for chamferless parts by using a very fast parallel robot called DELTA. The paper recalls the main features of the system in order to focus on the experimental evaluation of different search strategies employed to absorb misalignment between pieces, paying particular attention to the comparison of the mean search time of random and deterministic trajectories (such as random, Lissajous' and spiral trajectories) as a function of the clearance of parts. The experimental results obtained in this work show a better performance of the random strategy and stress the benefits of adopting a stochastic approach to design and implement open-loop control strategies for robotic assembly processes. © 1995.