The study includes two components：(1) cross-sectional (Study I), and (2) longitudinal treatment trial (Study II). The cross-sectional component will include all subjects initially recruited for the parent project. Genotyping characteristics will be compared with clinical status (i.e., recovered vs symptomatic). The treatment trial component (one) will include a subset of the subjects (n = 400) who remain significantly depressed. They will be randomly assigned to 8-weeks of treatment with either citalopram or paroxetine. With such a design, we wish to test the following hypotheses：

Ⅰ. Depressed patients with the short variant of the serotonin transporter (5HTTLPR) will respond faster and better to antidepressants compared to their counterparts with the long variant. Concurrently, patients with the 5-HTT Stin2 12/12 allele will also show better response as compared to those with the 10/12 allele.

Ⅱ. Depressed patients who are homozygous for deficient or less active CYP2D6 or CYP2C19 enzyme(s) will be more likely to show treatment emergent side effects compared to subjects with the wildtype alleles. Specifically, in Study II, CYP2D6 polymorphism will predict PAR but not CIT side effects and CYP2C19 polymorphism will be associated with CIT but not PAR side effects.

Despite remarkable progress in recent decades in modern psychopharmacotherapy, patients vary substantially in their response to antidepressants, ranging from total remission to complete treatment failure. Adverse effects, often bothersome and occasionally life-threatening, continue to represent significant challenges to patients and clinicians. Mechanisms responsible for such variability remain poorly understood. In addition, although less appreciated, substantial cross-ethnic variations in psychotropic responses often exist. Recent developments in the field of pharmacogenetics indicate that genetic factors may account for a large part of these differences in response. Specific genetic polymorphisms affecting the function of the serotonin (SERT) system has been postulated to predict the effect of antidepressants. Similarly, genetic mutations have been shown to exert a predominant influence on the expression of a number of drug-metabolizing enzymes, including most of the cytochrome P-450 enzymes (e.g., CYP2C19 and CYP3A4) that are responsible for the biotransformation of most antidepressants. Polymorphisms of genes controlling these enzymes have been found to be strongly associated with the propensity for various kinds of side effects. Capitalizing on these new developments, the proposed study will examine the predictive value of some of these genetic polymorphisms in 400 patients with DSM-IV major depression prospectively treated with citalopram (CIT). It is postulated that mutations affecting the function of SERT will predict responses to CIT, polymorphism of CYP2C19 will be associated with the side effect profiles and pharmacokinetics of CIT. The proposed study represents an extension and replication of a 5-year NIH/NIMH collaborative project that had designed and initiated in 2001 by the PI, which is currently ongoing at three sites in the U.S. ( "Ethnic Variations in Antidepressant Response" 1 R01 MH62421; 1R01MH626761R01MH62531, 07/01 - 06/06). In the original study, the inclusion of the two comparison groups, African Americans and Caucasians, whose genetic mutation patterns diverge significantly from each other, will allow us to examine how these differences affect their antidepressant response patterns and whether the associations are "replicable" across ethnicity. Results will be pooled with those derived from other sites, and will represent a rare opportunity to compare findings across Taiwanese, African American and Caucasian subjects with comparable diagnosis and treated with an identical protocol.

Inclusion Criteria:

1. self-identified as of Taiwanese ethnic background, and report that both of their parents and all four of their grandparents are members of the same ethnic group;
2. non-responders: have a 21-item HAM-D score of > 17; partial responders: have a 21-item HAM-D score between 8 and 15; responders: have a 21-item HAM-D score of < 7. Only the non-responder group will be included in Study II.
3. male or female, who, if of child-bearing potential, agrees to use effective contraception including the regular use of contraceptive pills, intra-uterine devises or abstinence;
4. age > 18;
5. capable of giving informed consent.

Exclusion Criteria:

1. Diagnosis of schizophrenia, schizophreniform disorder, schizoaffective disorder, schizotypal disorder, psychotic depression or bipolar disorders;
2. current drug or alcohol abuse or dependence or history of drug or alcohol abuse or dependence within the past 6 months;
3. unstable medical or neurological conditions that are likely to interfere with the treatment of depression;
4. history of allergy to antidepressants;
5. history of seizure disorder;
6. pregnancy;
7. active suicidal ideation or other safety issues determined by the clinician to not be suitable for inclusion in the study;