Normally use SQL Profiler GUI, but it's a bit of a pain.

Some links about scheduling tracing in SQL Server
http://www.mssqltips.com/tip.asp?tip=1715
http://www.novicksoftware.com/Articles/scripting-traces-for-performance-monitoring-on-sql-server-page3.htm
http://vyaskn.tripod.com/server_side_tracing_in_sql_server.htm
Of these, the first is straightforward and the last is a fairly comprehensive wrapper around the trace functions and includes data for all the event and column codes.
Using server tracing for performance evaluation

This describes the use of sys.traces: http://www.sqlservercentral.com/articles/Stairway+Series/72465/

1> select * FROM ::fn_trace_getinfo(default)
2> go
traceid property value
------- -------- -----
1       1        2
1       2        G:\Trace\20091023\trace_0344
1       3        5000
1       4        2009-10-23 03:49:03.497000000
1       5        1

Where property values are:

Each event can have one or more columns of data associated with it, these columns represent the values that are ultimately output (e.g. in SQL Profiler). Event and column defintions can be found here

1> select * FROM ::fn_trace_geteventinfo(1)
2> go
eventid columnid
------- --------
11 1    Textdata
11 2    BinaryData
11 3    DatabaseID
11 8    Hostname
11 9    Client process ID
11 10   Application ID
11 11   Login name
11 12   SPID
11 14   Start time
11 21   Event sub class
11 26   Server name
11 35   Database name
...
1> select distinct eventid FROM ::fn_trace_geteventinfo(1)
2> go
eventid
-------
11  RPC start
13  SQL batch start
14  Audit login
15  Audit logout
17  Existing connection - detects activity from before trace started
53  Cursor Open
70  Cursor prepare
71  Prepare SQL
72  Exec prepared SQL
74  Cursor execute
77  Cursor unprepare
78  Cursor close
100 RPC output parameter

Column definitions are here

1> select * FROM ::fn_trace_getfilterinfo(1)
2> go

columnid logical_operator comparison_operator value
-------- ---------------- ------------------- -----
3        0                0                   17
3        1                0                   16
10       0                7                   SQL Profiler

A trace is created by executing sp_trace_create, e.g.

declare @TraceID int
declare @maxfilesize bigint -- The maximum trace file size in megabytes
set @maxfilesize = 5
exec @rc = sp_trace_create @TraceID output, 0, N'InsertFileNameHere', @maxfilesize, NULL
if (@rc != 0) goto error
goto finish
error:
select ErrorCode=@rc
:finish
go

Note that traces created this way are initially stopped and do not have any events or filters associated with them.
Use sp_trace_setevent to add events and columns, e.g.

-- 11 = RPC start
exec sp_trace_setevent @TraceID, 11, 1,  @on    -- Textdata
exec sp_trace_setevent @TraceID, 11, 2,  @on    -- Binary data
exec sp_trace_setevent @TraceID, 11, 3,  @on    -- Database ID
exec sp_trace_setevent @TraceID, 11, 8,  @on    -- Hostname
exec sp_trace_setevent @TraceID, 11, 9,  @on    -- Client ID
exec sp_trace_setevent @TraceID, 11, 10, @on    -- Application ID
exec sp_trace_setevent @TraceID, 11, 11, @on    -- login name
exec sp_trace_setevent @TraceID, 11, 12, @on    -- SPID
exec sp_trace_setevent @TraceID, 11, 14, @on    -- Start time
exec sp_trace_setevent @TraceID, 11, 21, @on    -- Event sub class
exec sp_trace_setevent @TraceID, 11, 26, @on    -- Server
exec sp_trace_setevent @TraceID, 11, 35, @on    -- Database name

Use sp_trace_setfilter to add filters, e.g.

-- Database ID = 16
exec sp_trace_setfilter @TraceID, 3, 1, 0, 16
-- Exclude SQL Profiler
exec sp_trace_setfilter @TraceID, 10, 0, 7, N'SQL Profiler%'

Use sp_trace_setstatus to start, stop and close the trace, e.g.

exec sp_trace_setstatus @traceid=1, @status=1 -- start trace
exec sp_trace_setstatus @traceid=1, @status=0 -- stop/pause trace
exec sp_trace_setstatus @traceid=1, @status=2 -- close and delete trace