Abstract We report on the results of a systematic study of X-ray flares from low-mass young stellar objects, using two deep exposure Chandra observations of the main region of the $\rho$ Ophiuchi star-forming cloud. From 195 X-ray sources, including class I–III sources and some young brown dwarfs, we detected a total of 71 X-ray flares. Most of the flares have the typical profile of solar and stellar flares, fast rise and slow decay, while some bright flares show unusually long rise timescales. We derived the time-averaged temperature ( $\langle kT\rangle$ ), luminosity ( $\langle L_{\mathrm{X}}\rangle$ ), rise and decay timescales ( $\tau_{\mathrm{r}}$ and $\tau_{\mathrm{d}}$ ) of the flares, finding that (1) class I–II sources tend to have a high $\langle kT\rangle$ , which sometimes exceeds 5 keV, (2) the distribution of $\langle L_{\mathrm{X}}\rangle$ during flares is nearly the same for all classes from $\sim 10^{29.5}$ to $\sim10^{31.5} \,\mathrm{erg} \,\mathrm{s}^{-1}$ , although there is a marginal hint of a higher $\langle L_{\mathrm{X}}\rangle$ distribution for class I than class II–III, and (3) positive and negative log-linear correlations are found between $\tau_{\mathrm{r}}$ and $\tau_{\mathrm{d}}$ , and $\langle kT\rangle$ and $\tau_{\mathrm{r}}$ . In order to explain these relations, we used the framework of magnetic reconnection model with heat conduction and chromospheric evaporation to formulate the observational parameters ( $\tau_{\mathrm{r}}$ , $\tau_{\mathrm{d}}$ , and $\langle kT\rangle$ ) as a function of the pre-flare (coronal) electronic density ( $n_{\mathrm{c}}$ ), the half-length of the reconnected magnetic loop ( $L$ ), and magnetic field strength ( $B$ ). The observed correlations are well reproduced if loop lengths are nearly the same for all classes, regardless of the existence of an accretion disk. The estimated loop length is almost comparable to the typical stellar radius of these objects ( $10^{10} \hbox{--} 10^{11} \,\mathrm{cm}$ ), which indicates that the observed flares are triggered by solar-type loops, rather than larger ones ( $\sim 10^{12} \,\mathrm{cm}$ ) connecting the star with its inner accretion disk. The higher $\langle kT\rangle$ observed for class I sources may be explained by a slightly higher magnetic field strength ( $\approx 500 \,\mathrm{G}$ ) than for class II–III sources (200–300 G).