This is a known problem. See here. There are some possible workarounds though. If you don't mind about PDFs you can use the "export citations" button on the sciencedirect site - if you export in RIS format Zotero should automatically intercept the export and save the items (but without PDFs). Make sure that "use Zotero for downloaded RIS/refer files" is checked in the prefs.

If you need the PDFs you can use a mass downloading tool such as "DownThemAll" (a firefox addon) to save them to your computer, then import from there into Zotero and either manually match up the PDFs to the items from the "export citations" method (a real pain given the unhelpfully named PDFs - you might be quicker saving each article individually) or use "retrieve metadata for the PDFs to automatically build items around the PDFs. Please note that you might well get temporarily blocked from google scholar if you send too many "retrieve metadata" requests. I'm afraid no one knows what the exact limit is as google hasn't published this, but I would be cautious about doing more than 10-15 articles in a short space of time. Don't try this method if temporarily losing access to google scholar (usually for 24 hours) would be a major problem for you.

I don't know about the timescale on fixing the translator for ScienceDirect as I am not a developer. From the number of times the translator has broken over the years due to changes at ScienceDirect's end it seems that this is an exceptionally difficult site to support (although I agree that good support is pretty much essential for users in the natural sciences).

ajlyon - no, unfortunately not:

http://www.sciencedirect.com/science/journal/02613794

[JavaScript Error: "TypeError: newDoc.evaluate("//a[contains(@class, \"icon_exportarticlesci_dir\")]", newDoc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext() is null" {file: "file:///home/sebastian/.mozilla/firefox/wkxw8okh.default/zotero/translators/ScienceDirect.js" line: 0}]

Here is the debug output:
(4)(+0003815): Translate: Parsing code for ScienceDirect

(5)(+0000019): Translate: running handler 0 for select

(3)(+0003609): created hidden browser (1)

(3)(+0000001): loading http://www.sciencedirect.com.turing.library.northwestern.edu/science?_ob=ArticleURL&_udi=B6V9P-4YNB508-B&_user=965532&_coverDate=09%2F30%2F2010&_rdoc=2&_fmt=high&_orig=browse&_srch=doc-info(%23toc%235904%232010%23999709996%232212738%23FLA%23display%23Volume)&_cdi=5904&_sort=d&_docanchor=&_ct=28&_acct=C000049540&_version=1&_urlVersion=0&_userid=965532&md5=557c386e3167343d3f26d9e966311075

(3)(+0000992): http://www.sciencedirect.com.turing.library.northwestern.edu/science?_ob=ArticleURL&_udi=B6V9P-4YNB508-B&_user=965532&_coverDate=09%2F30%2F2010&_rdoc=2&_fmt=high&_orig=browse&_srch=doc-info(%23toc%235904%232010%23999709996%232212738%23FLA%23display%23Volume)&_cdi=5904&_sort=d&_docanchor=&_ct=28&_acct=C000049540&_version=1&_urlVersion=0&_userid=965532&md5=557c386e3167343d3f26d9e966311075 has been loaded

(3)(+0000006): deleted hidden browser

(2)(+0000001): Translate: Translation using ScienceDirect failed:
message => newDoc.evaluate("//a[contains(@class, \"icon_exportarticlesci_dir\")]", newDoc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext() is null
fileName => chrome://zotero/content/xpcom/translate.js
lineNumber => 899
stack => ([object XPCNativeWrapper],[object Object])@chrome://zotero/content/xpcom/translate.js:899
([object XPCNativeWrapper])@chrome://zotero/content/xpcom/translate.js:928

name => TypeError
url => http://www.sciencedirect.com/science/journal/02613794
downloadAssociatedFiles => true
automaticSnapshots => true

(3)(+0000037): [removed the list of my extensions - if it turns out to be one of them I'll figure it out]

(5)(+0000002): Translate: running handler 0 for done

(4)(+0005174): Registering observer for [collection,search,share,group,bucket] in notifier with hash tb'

(5)(+0000002): SELECT itemTypeID AS id, typeName AS name, custom FROM itemTypesCombined WHERE display=2

(5)(+0000001): SELECT itemTypeID AS id, typeName AS name, custom FROM itemTypesCombined WHERE display=1

I haven't worked on it much, but as far as I know the general problem with ScienceDirect is that people seem to get different versions of the site based on all sorts of factors, which makes it very hard to create a version that works for everyone—that's more likely why it wasn't working for adamsmith. But ideally we can account for all modes.