A unidirectional high current pulse spark with a very fast rise-time, ensuring a rapid and complete transfer of energy to the sample, was used as the sampling system for the anal. of C steels and highly alloyed steels with the same operating conditions. The sparking operating conditions were optimized and a restrictive path was designed to decrease the quantity of eroded material reaching the plasma, to prevent deposition of material in the torch injector, to minimize sampling cone blockage and drift effects. Spark ablation sampling efficiency and effectiveness of the restrictive path were evaluated. To compensate for differences in the amt. of material ablated or for a variation in drift, 57Fe and 55Mn were used as internal stds. The calibration procedure was applied to the anal. of the elements Al, B, Co, Cu, Mn, Nb, P, Si, and V, present in the following certified ref. materials: BCS (Bureau of Analyzed Samples) SS-456 to SS-460 (residual series); CRMs (European Committee for Iron and Steel Standardization) No 285-2 (Maraging steel), No 292-1 (Nb stabilized steel), No 295-1 (highly alloyed steel), and No 296-1 (jethete steel). When plotting intensity ratios (IX/IIS) vs. concn. ratios (CX/CIS) linear calibration curves over the entire range of tested concns., with correlation coeffs. better than 0.999, were obtained. Detn. limits 0.03% with relative std. deviation values <3%. For the elements As, Sn, Ti, W and Zr, only present in one or two of the Std. Materials, the sensitivity was also evaluated. Also, the possibility of obtaining reproducible transient signals from sparking periods of only a few seconds was demonstrated.