The study evaluates two alternative seismic intensity measures (IMs) that reduce the collapse capacity dispersion of inelastic non-degrading single-degree-of-freedom (SDOF) systems vulnerable to the P-delta effect. This dispersion of collapse capacity is caused by record-to-record variability, which refers to frequency content variation of the ground motions used in the dynamic analyses. This reduction (of dispersion) is achieved utilizing efficient elastic pseudo-spectral acceleration based IMs. The first set of evaluated IMs is based on the spectral pseudo-acceleration averaged in a certain period interval between the structural period and an elongated period. The ''optimal'' lower bound of the period interval corresponds to the structural period of vibration, since naturally in an SDOF system no higher modes effects do exist. The ''optimal'' upper bound of the period interval for averaging, referred to as elongated period, is found to be 1.6 times the system period. The second IM considered in the study is the 5 % damped spectral pseudo-acceleration at the system period in the presence of gravity loads, which is a single target IM. The most widely accepted IM, the 5 % damped pseudo-spectral acceleration at the system period without P-delta, serves as the benchmark IM. The results show that both proposed IMs lead to a reduction of the collapse capacity dispersion compared to the benchmark IM outcomes. The IM based on the averaged spectral acceleration of the ''optimal'' period interval is more efficient up to a negative post-yield stiffness ratio of 0.45, while the single target IM based on the system period in the presence of gravity loads is superior for extreme negative post-yield stiffness ratios larger than 0.45. Additionally, the sufficiency and the scaling robustness property of the considered IMs with respect to the natural logarithm of the record-dependent individual collapse capacities is discussed for a wide range of structural configurations.