The impact of three altimeter sampling patterns (Geosat, ERS-1 and Topex-Poseidon) on the estimation of mesoscale eddy sea surface height anomalies is investigated in detail by making use of a regional model of the Gulf Stream and of an objective analysis. It is shown that, in a domain-averaged statistical sense, track sequencing does not appear to be of importance in the eddy field recovery. On the other hand, significant differences appear when one starts to look at specific processes such as rings. Qualitatively speaking, the manner in which a ring's signature is represented depends strongly upon the altimeter orbit choice. A higher space sampling such as ERS-1 is preferable in order to keep track of a ring, but to the detriment of its amplitude. A low space sampling (or high time sampling) such as Topex-Poseidon does not provide enough information and the ring's signature vanishes from the resulting maps.