The primary question addressed by this prevention study is to explore if improving pain and disability reduces episodes of Major Depression among seniors with knee osteoarthritis (OA) and mild depressive symptoms.

In this Sequential Multiple Assignment Randomized (SMART) Trial, the investigators care comparing Cognitive Behavioral Therapy for Pain (CBT-P) with EXERCISE (knee-specific Physical Therapy) delivered individually and sequentially. A subset of participants will receive enhanced care as usual in which their primary care physicians will receive information about pharmacological management of knee osteoarthritis. The active phase of the study may be up to 16 weeks, and the follow-up period lasts 12 months.

Stage 1, Specific Aim 1: To test if Cognitive Behavioral Therapy for Pain (CBT-P) results in more improvement in pain and disability than those receiving knee-specific Physical Therapy (EXERCISE).

Stage 1, Hypothesis 1: At week 8, subjects randomized to CBT-P will have more improvement in pain and disability than those randomized to EXERCISE.

Stage 2, Specific Aim 2: For non-responders to Stage 1, to explore which sequence of interventions leads to greater improvement in pain and disability.

Stage 2, Hypothesis 2: Compared to the other sequenced interventions (see Figure), subjects randomized to CBT-P and then EXERCISE will have the most improvement in pain and disability.

Follow-up, Specific Aim 3: To explore if improvement in pain and disability is associated with incident major depressive episode (MDE) over 12 months.

Follow-up, Hypothesis 3: More pain and disability improvement will be associated with lower rates of incident (MDE) over 12 months.

Exploratory AIM 1: To compare "legacy" assessments of functioning and psychological health with Computer Adaptive Testing (CAT) assessments of these domains for level of agreement.

Exploratory AIM 2: To assess candidate genetic polymorphisms, peripheral mRNA biosignatures, and candidate cytokine and growth factor levels using both multiplex assays and individual ELISAs. The investigators are exploring if these biomarkers are associated with pain improvement and risk for developing MDE.

Inclusion Criteria:

1. >/= age 60
2. meets accepted clinical criteria for knee OA based on the American College of Rheumatology 1986 clinical criteria guidelines.
3. Western Ontario and McMaster University Arthritis Index (WOMAC) pain subscale score in the range of 7-15 (to minimize including subjects with knee OA so severe they may be better served by arthroplasty)
4. PHQ-9 scores 1-9, with at least one of the cardinal symptoms of depression (low mood or anhedonia) endorsed.
5. Modified Mini Mental State (3MS) Examination >/= 80.
6. Has or is willing to establish care with a personal physician prior to any experimental procedures.

Exclusion Criteria:

1. Major Depressive Episode or anxiety disorder within the past 1 year
2. Currently taking an antidepressant
3. History of bipolar or schizophrenia
4. Drug or alcohol use disorder within the past 12 months
5. Receiving knee-related workers compensation or involved in knee pain-related litigation.
6. Currently taking an anti-anxiety medicine > 4 times/week for the past 4 weeks.