ELDORA observations of the Garden City tornadic and Hays nontornadic supercells during VORTEX95 field experiment are presented. The kinematic features of both storms are obtained from dual-Doppler analysis of airborne Doppler radar. Thermodynamic fields are retrieved by pressure-buoyancy retrievals. Detailed analyses of the Hays supercell shows at least 5 low-level mesocyclones during the data collection period of ELDORA, while the Garden City tornadic supercell only produced one low-level mesocyclone. All 5 mesocyclones in the Hays nontornadic supercell are carefully analyzed and the strongest one is compared with the Garden City tornadic mesocyclone. It is found that both tornadic and nontornadic mesocyclones could go through an identical evolution process and show strikingly similar characteristics in various aspects such as vertical vorticity, updraft-downdraft pattern and swirl ratio. The similar kinematic structure between tornadic and nontornadic mesocyclones could produce serious challenges for detecting tornadogenesis using the WSR-88D algorithms. Why did the strongest low-level mesocyclone in the Hays supercell failed to produce a tornado? It is believed that another newly formed vortex, which developed along the rear flank gust front is responsible. The new vortex intercepted the warm inflow into the strongest low-level mesocyclone and cause its demise. All other mesocyclones in the Hays supercell failed to produced a tornado because their swirl ratios were not optimal. Gust front evolutions for tornadic Garden City mesocyclone and nontornadic Hays mesocyclone are also analyzed and some hypotheses for tornadogenesis or failure are examined.