We discuss the possibility of using fast radio bursts (FRBs), if cosmological, as a viable cosmic probe. We find out that the contribution of the host galaxies to the detected dispersion measures can be inapparent for the FRBs not from galaxy centers or star forming regions. The inhomogeneity of the intergalactic medium (IGM), however, causes significant deviation of the dispersion measure from that predicted in the simplified homogeneous IGM model for individual event. Fortunately, with sufficient FRBs along different sightlines but within a very narrow redshift interval (e.g., $Δ z ∼ 0.05$ or $Δ z ∼ 0.05(1+z)$), the mean from averaging observed dispersion measures does not suffer such a problem and hence may be used as a cosmic probe. We show that in the optimistic case (e.g., tens FRBs in each $Δ z$ have been measured; the most distant FRBs were at redshift $≥ 3$; the host galaxies and the FRB sources contribute little to the detected dispersion measures) and with all the uncertainties (i.e. the inhomogeneity of the IGM, the contribution and uncertainty of host galaxies as well as the evolution and error of $f_{\rm IGM}$) considered, FRBs could help constrain the equation of state of dark energy. ; Comment: 5 pages including 2 figures. The version published in PRD, minor revision