The performance of different implementations of the ISO transport protocol class 4 on top of a connectionless link-layer protocol in the presence of transmission errors is studied by means of simulation. A main goal of the study is to find robust protocol versions yielding good performance for all possible combinations of implementations. The performance measure investigated is the throughput that can be achieved in transferring files between two stations over a local area or long-distance network. The results are compared to those obtained for the ISO transport protocol class 2 on top of a connection-oriented link-layer protocol. It is assumed that data and acknowledgement frames could be lost or corrupted by transmission errors. The results show that, at the sender side, retransmit all always yields higher performance than retransmit first, and, at the receiver side, store results in higher performance than discard.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>