Water found as active leaks and isolated pools in the Weeks Island, Jefferson Island, and Belle Isle salt mines of south Louisiana has Δ18O values ranging from -4 to +11.5%0 and ΔD values from -2.3 to -53%0. One sample from Weeks Island and one from Jefferson Island are isotopically similar to local surface waters and are clearly of meteoric origin. All other samples are too enriched in 18O to be meteoric waters. In the Weeks Island mine the isotopic data define a linear array given by ΔD=3.0Δ18O-40.1. Active leaks define the positive end of this array. Isolated pools are interpreted as inactive leaks with initial Δ18O and ΔD values of +9.1¿0.5%0 and -11%0¿7%0, which have subsequently exchanged with water vapor in the mine air to produce the linear array of Δ values. The water derived from active leaks in these three mines is too enriched in 18O and too depleted in D to be connate ocean water or evaporite connate water trapped in the salt. The isotopic composition of water derived from the dehydration of gypsum is probably dissimilar to that of the active leaks. It is unlikely that the water has originated from the dehydration of gypsum. It is also unlikely that isotopic exchange with anhydrite is responsible for the observed 18O enrichments. Nonmeteroric water from the active leaks displays the type of 18O enrichments characteristic of saline formation waters, where water exchanges isotopically with calcite and clay minerals. It is concluded that the nonmeteoric waters are formation waters which have become incorporated in the salt. From the observed 18O enrichment it is calculated that the formation waters were incorporated during diapiric rise of the salt at a depth of 3--4 km and have been trapped within the salt for 10--13 m.y. Large volumes of salt within salt domes are domes are not naturally penetrated by meteoric groundwaters but can contain limited amounts of trapped formation water which will have to be contended with in any attempt to store radioactive wastes in salt domes.