2 lines in which I basically give both images a name. The reason the watermark is a png is I want to have the ability to use alpha channels.
<cfimage source="img_2566.jpg" name="image" / >
<cfimage source="watermark.png" name="watermark" / >

Next paste the watermark onto the image at 0x and 0y. I cheated and created the images the same size, but you could easily read the image info get the size then calculate where you want the watermark to be.
<cfset ImagePaste(image, watermark, 0, 0) / >

The last line writes the image to the browser in the format of a jpeg.
<cfimage source="#image#" action="writeToBrowser" format="jpg" / >

The Final Code should look like this
<cfimage source="img_2566.jpg" name="image" / >
<cfimage source="watermark.png" name="watermark" / >
<cfset ImagePaste(image, watermark, 0, 0) />
<cfimage source="#image#" action="writeToBrowser" format="jpg" / >

More information can be seen here:
http://www.damonledet.com/cfwatermark/