We show that the Abelian topological mass mechanism in four dimensions, described by the CremmerSherk action, can be obtained from dimensional reduction in five dimensions. Starting from a gauge invariant action in five dimensions, where the dual equivalence between a massless vector field and a massless secondrank antisymmetric field is established, the dimensional reduction is performed keeping only one massive mode. ͓S0556-2821͑98͒03324-4͔ PACS number͑s͒: 11.10. Lm, 11.25.Mj Several alternatives of the Higgs mechanism, based on the coupling of vector fields with antisymmetric fields through topological terms, have been developed in the last years. In particular, the Abelian Cremmer-Sherk theory ͓1͔ has been studied extensively ͓2͔ as an appropriate model for this proposal. Its non-Abelian extension is a Freedman-Townsend theory ͓3͔ that can be derived using the self-interaction mechanism ͓4͔. Also, other attempts to search a non-Abelian generalization have been formulated ͓5͔; however, serious problems with renormalizability in all these non-Abelian generalizations has been pointed out in Ref.͓6͔. An interesting aspect is the fact that the Cremmer-Sherk theory is related by duality ͓7,4͔ with the Kalb-Ramond theory ͓8͔, where the latter can be obtained by dimensional reduction of the second-rank antisymmetric theory in five dimensions keeping only one massive mode ͓9,10͔. It is worth recalling that the Kalb-Ramond theory provides mass in a nontopological way and can be understood as the resulting theory after condensation of magnetic monopoles in four dimensions QED ͓11͔. In this Brief Report, we show that it is possible to obtain the Cremmer-Sherk theory in four dimensions by dimensional reduction, despite Barcelos-Neto's claim that the Cremmer-Sherk theory cannot come from dimensional reduction of any gauge theory in five dimensions ͓10͔.Let us describe briefly the dual equivalence between the Cremmer-Sherk and Kalb-Ramond theories in four dimensions. The Cremmer-Sherk action, which provides mass to spin-1 fields without spoil gauge invariance, is written down asmn Ϫ 1 12 H mnp H mnp Ϫ 1 4 ⑀ mnpq B mn F pqͬ , ͑1͒ where F mn ϵץ m A n Ϫץ n A m and H mnp ϵץ m B np ϩץ n B pm ϩץ p B mn are the field strengths associated with the A m and B mn fields. This action is invariant under gauge transformations: ␦A m ϭץ m and ␦B mn ϭץ m n Ϫץ n m . We observe that this action has global symmetries, for instance, A m →A m Ϫ⑀ m , with ⑀ m the global parameter. A useful way to obtainthe dual theory relies on gauging this local symmetry ͓12͔ by introducing an antisymmetric field a mn , such that the field strength F mn is modified by F mn ϵץ m A n Ϫץ n A m ϩ 1 2 (a mn Ϫa nm ) and adding a BF term, which assures the nonpropagation of a mn . Then, we have the following action:and we now have new gauge symmetries, given by ␦A m ϭϪ⑀ m(x) , ␦a m ϭץ m ␣, and ␦a mn ϭץ m ⑀ n Ϫץ n ⑀ m , which allow us to fix the gauge A m ϭ0. After fixing this gauge, the action becomes1 4 a mn a mn Ϫ 1 12 H mnp H mnp Ϫ 1 4 ⑀...