Thermo-mechanical stress in MCM-D substrate is an important reliability and fabrication issue. The difference in coefficient of thermal expansion(CTE) between substrate, polymer, and metal leads to complicated stress fields in multilevel interconnect structures. This study uses materials set representative of typical MCM-D technology to monitor the stress level in the thin film layers of polymer. A Kapton PI film is bonded to a substrate with heat and pressure using a thin layer of Ultem 1000 thermoplastic or Coverlay thermosetting adhesive. Mechanical test structures consist of single or multilayer polymer films are fabricated on silicon wafers. The substrate deflection caused by composite stresses due to fabrication and thermal cycling is determined by a curvature measurement technique. Initial stress or bending is intrinsic but becomes extrinsic upon thermal cycling. The composite stress or bending in multilayer structures is due to a contribution from each individual layer. In a thermo-mechanical stress viewpoint, Ultem 1000 thermoplastic and the epoxy thermosetting adhesive exhibit quite different behaviors. While the epoxy thermosetting adhesive exhibits an ideal elastic behavior predicted from the elastic analysis, the total thermo-mechanical stress tends to decrease when Ultem thermoplastic is used as an adhesive. It seems that this relaxation effect of Ultem thermoplastic is due to the self-relaxation nature of thermoplastic materials themselves, while the ideal elastic behavior of the epoxy thermosetting is due to the heavily crosslinked structure of thermosetting materials. A simpler model which predicts a stress contribution from each individual layer during MCM-D substrate fabrication is proposed and verified.