Let $P = (\Omega, \mathcal{F}, \mathbb{P})$ be a D1159: Probability space such that

(i)	$\mathcal{G} \subseteq \mathcal{F}$ is a D470: Subsigma-algebra of $\mathcal{F}$ on $\Omega$
(ii)	$X_j$ be a D202: Random variable on $P$ for each $j \in J$
(iii)	$\sigma_{\text{pullback}} \langle X_j \rangle$ is the D1730: Pullback sigma-algebra of $X_j$ on $P$ for each $j \in J$

Then $\{ X_j \}_{j \in J}$ is a conditionally independent random collection on $P$ given $\mathcal{G}$ if and only if \begin{equation} \forall \, N \in \{ 1, 2, 3, \ldots \} : \forall \, j_1, \, \ldots, \, j_N \in J \left[ E_{j_1} \in \sigma_{\text{pullback}} \langle X_{j_1} \rangle, \, \ldots, \, E_{j_N} \in \sigma_{\text{pullback}} \langle X_{j_N} \rangle \quad \implies \quad \mathbb{P} \left( \bigcap_{n = 1}^N E_{j_n} \mid \mathcal{G} \right) \overset{a.s.}{=} \prod_{n = 1}^N \mathbb{P}(E_{j_n} \mid \mathcal{G}) \right] \end{equation}