Since transition metal catalysis allows simultaneous formation of more than one bond in a single step operation with high selectivity, the exploitation of a model compound for nitrogen-containing polycyclic biologically active natural products has been a worthwhile contribution in synthetic and medicinal chemistry. Intramolecular hydroamidation of acetylenes, carbonylation of 2-alkylbenzenylamines, 1-aryl-3-hexen-1,5-diynes initiated by methoxide addition, fluoroarenes and nitriles via 1,2-arynes, nitroarylstannanes, radical dearomatization of benzene and hydrothermal reaction of o-phenylaniline, ullamann cross-coupling of 1-bromo-2-nitroarenes and 1,3dipolarcycloaddition of nonstabilized azomethine ylide and photocycloaddition of phthalimide anion to alkenes have been used to synthesize benzazepinone, phenanthridine and benzophenathridinone derivatives. However, these synthetic routes are often complicated and limited to only some substituents. Recently, we reported various cycloisomerization reactions with different unsaturated systems catalyzed by Au, Pt, Pd, and Rh. Our continued interest in the synthesis of polycyclic systems prompted us to develop the Rhcatalyzed cyclization of o-alkynylbenzaldehydes having a nitrile tether that produced tricyclic benzazepinones and benzophenanthridine derivatives via [3+2] and [4+2] cycloaddition via pyrylium intermediates as shown in Scheme 1. An initial study was tested with o-alkynylbenzaldehyde 1a in the presence of various late transition metal catalysts (Table 1). PtCl2 as a catalyst in refluxing 1,4-dioxane for 12 h converted 1a to give a mixture of products, which were separated by column chromatography to give 2,3-dihydrobenzo[e]cyclopenta[b]azepin-5(1H)-one (3a) in 50% yield along with the hydrated product 2a in 30% yield (entry 1). When this reaction was carried out in dry 1,4-dioxane in the presence of 4 A molecular sieves, the reaction was dramatically accelerated to give the product 3a in 59% yield still along with 2a in 10% yield (entry 2). We presumed that a trace amount of water in the reaction medium or in solvent might generate a nucleophile which could undergo hydration to the metal-activated triple bond followed by tautomerization to the corresponding ketoaldehyde 2a. It was observed that gold(III) bromide in 1,2-dichloroethane (EDC) hydrated 1a to afford 2a even at room temperature exclusively (entry 3). This might be understood by the high reactivity of gold catalyst or the poor reactivity of the nitrile group toward the pyrylium intermediate A. We tried the same reaction with different catalysts, such as AuCl3, AuCl(PPh3) with AgSbF6, and AgSbF6 itself, under dry reaction condtions, but led to hydration (entries 4-6). In fact, Zhu et al. reported that o-alkynylbenzaldehydes were hydrated in the presence of gold(+3) catalysts and trifluoroacetic acid to afford the corresponding ketoaldehyde. Finally, we found [Rh(COD)Cl]2 as an optimal catalyst for the present [3+2] cyclization of pyrylium intermediate, generated in situ with alkynophilic metal cations, with a pendent nitrile. Thus, when the substrate 1a in the presence of [Rh(COD)Cl]2 (10 mol%) in dry xylene containing 4 A molecular sieves was heated at 110 C for 12 h, the product 3a was isolated in 71% yield along with the ketoaldehyde 2a in 10% yield. With this promising result, we further examined this [3+2] cycloaddition with structural alteration in the tether (eq. 1-2). This paper is dedicated to Professor Sang Chul Shim at Kyungpook National University on the occasion of his honorable retirement. Scheme 1 Table 1. Optimization of intramolecular [3 + 2] cyclization-cycloaddition under various reaction conditions with 1a