In addition to the 2'-azido analogue of (I)n x (C)n, (dIn3)n x (C)n, we have found two other (I)n x (C)n analogues, (dIfl)n x (C)n and (dIcl)n x (C)n, in which the 2'-hydroxyls of the (I)n strand are replaced by either fluorine or chlorine, to be highly effective in inducing interferon. This contrasted with the lack of interferon-inducing activity noted for various other 2'-halogeno analogues of (I)n x (C)n and (A)n x (U)n, i.e. (I)n x (dCcl)n, (dAfl)n x (U)n, (dAcl)n x (U)n, (A)n x (dUfl)n and (A)n x (dUcl)n. In most assay systems, viz. primary rabbit kidney cells, human diploid fibroblasts, HeLa cells, interferon-primed mouse L-929 cells, and intact rabbits, (dIfl)n x (C)n and (dIcl)n x (C)n induced interferon levels that were comparable to those induced by (I)n x (C)n. There was one particular system (L-929 cells treated with DEAE-dextran), however, in which (dIfl)n x (C)n and (dIcl)n x (C)n, unlike (I)n x (C)n, failed to stimulate interferon production. As monitored by both radiochemical and biological means, (dIfl)n x (C)n and, to a lesser extent, (dIcl)n x (C)n were more resistant to degradation by ribonuclease A, T1 and human serum nucleases than was (I)n x (C)n. In their reactivity towards antibodies to double-stranded RNA (dIfl)n x (C)n and (dIcl)n x (C)n conformed more closely to (I)n x (C)n than did other 2'-substituted (e.g. 2'-O-methyl or 2'-O-ethyl) analogues of (I)n x (C)n. The high interferon-inducing potency of (dIfl)n x (C)n and (dIcl)n x (C)n has both theoretical and practical implications. While our findings suggest that (dIfl)n x (C)n and (dIcl)n x (C)n should be further explored for their therapeutic potentials, they also strengthen the notion that the interferon-inducing capacity, and possibly other biological functions of double-stranded RNAs is dependent on the recognition of the overall conformation of the polynucleotide rather than on the binding of specific functional groups such as the 2'-hydroxyl group.