A professor in graduate school would quip: "Partialize and Prioritize." In other words, try to break problems down into manageable sections, and take them one at a time. But which one first? I've heard different schools of thought: either start with a relatively easy, achievable goal to increase self-efficacy and help people feel like therapy is helping (particularly if you don't yet have a rapport with them), or start with the most urgent concern. However, there are certainly other factors that influence this decision - for example, an insurance company that recently told my colleague they'd pay for a client to receive treatment for one problem, but not for another.
DBT actually offers a sequence of "primary behavioral targets," starting with risk issues (suicide, self-harm), then "therapy-interfering behaviors," then behaviors that interfere with quality of life (dysfunctional behaviors of all kinds, including substance abuse, impulsivity, and poor judgment), then increasing coping/life skills, then decreasing other symptoms.
I tend to agree that you have to start with imminent risk, and anything that is impeding or preventing treatment. I tend to include substance use in the latter category if people can't address emotional issues because they are under the influence and pretty much numb. However, with multiple problems, all of which are somewhat risky, I try to figure out whether one problems seems to be driving others, and start there.
However, like the metaphorical chicken and egg, it's rarely easy or clear to determine causality. Sometimes one problem emerged first (depression, and then an eating disorder, for example), but people often aren't aware of a sequence. Starting with whatever feels more important to the client is a good alternative (or even primary) strategy. Then, if another problem begins to prevent further progress on the top-priority issue, it becomes clear how they're related, and how to begin untangling them. For example, a client entering treatment for an eating disorder, who is totally unmotivated to change their behavior (and unresponsive to motivational interviewing) because depression has sapped all motivation about anything. Unless we treat the depression, we're unlikely to get far with anything else.
How do you "partialize and prioritize"? How do you make decisions about what to focus on when clients present with comorbid problems?