The 1(π,π*) vertical transition of ethylene is studied by means of an MRD-CI all-valence-electron treatment employing a relatively large AO basis containing two separate cartesian d polarization sets located at each of the carbon atoms; 84 AO's are present in the basis and the t,.velve valence electrons Of C 2H 4 are distributed among a set of 80 MO's or NO's in the CI calculations. Results are obtained for four different pairs of diffuse d exponents and it is found that (a) (ψ|Σ x i 2 |ψ) values for the 1(π,π*) state are relatively insensitive to the choice of such exponents but that (b) nearly optimal values thereof are 0.030 and 0.022 respectively. The best estimates obtained from the present study for (ψ|Σ x i 2 |ψ) and the vertical energy difference and f value for the 1 (π,π*) transition of ethylene are 20 au, 7.95 eV and 0.34 respectively. The former value is 6 au lower than what is obtained with the best Cl treatment reported to date for an AO basis with a single carbon d primitive (exponent 0.75 The present vertical ΔE value is 0.16 eV lower than in the Brooks-Schaefer study, but it still lies 0.3 eV above the location of the corresponding Franck-Condon absorption maximum, thereby continuing to indicate a significant amount of non-verticality for this transition, as has also been concluded on the basis of independent calculations of the vibrational intensity distribution in this band system. The computed f value agrees with the corresponding measured result to within experimental error. In view of the large oscillator strength it is understandable that experimentalists have traditionally looked upon this state as an essentially intra-valence species; nevertheless the calculations indicate, that it does contain a significant amount of diffuse character, as exemplified by the fact that exclusion of all diffused AO's from the AO basis leads to an increase in the 1 (π,π* ) energy of 0.5 eV over and above the lowest value obtained with an optimized set of Rydberg functions. Finally evidence is reported which strongly supports the view that the configuration selection and energy extrapolation techniques employed in this work and previously (with an AO basis less than half as large as in the present case) lead to quite accurate estimations of both property and energy results at the limit of zero selection threshold.